This topic provides answers to some frequently asked questions about Elastic Compute Service (ECS) instances.
FAQ about purchasing instances
FAQ about enterprise-level instances
What are enterprise-level instances? What are entry-level instances?
What are the differences between enterprise-level and entry-level instances?
In what business scenarios do I need to purchase enterprise-level instances?
How is the network performance of enterprise-level instances?
Which cloud disk categories do enterprise-level instances support?
What are the limits on changing the instance types of enterprise-level instances?
Can I upgrade an entry-level instance to an enterprise-level instance?
FAQ about persistent memory-optimized instances
FAQ about universal instances
FAQ about e instances
FAQ about GPU-accelerated instances
FAQ about ECS bare metal instances
FAQ about Super Computing Cluster (SCC)
FAQ about preemptible instances
FAQ about purchasing preemptible instances
FAQ about changing preemptible instances
FAQ about releasing preemptible instances
Can I cancel or reschedule the automatic release of a preemptible instance?
I do not have overdue payments in my account. Why have my preemptible instances been released?
Am I notified when my preemptible instances are about to be released? How am I notified?
Can the data of a preemptible instance be automatically retained when the instance is released?
What do I do if I cannot restart a preemptible instance that was stopped in economical mode?
How do I retain the public IP address of a preemptible instance when the instance is released?
FAQ about preemptible instance billing
How is a preemptible instance billed if its lifespan is shorter than 1 hour?
To what resources are the prices of preemptible instances applicable?
May the spot price of a preemptible instance exceed the pay-as-you-go price?
Am I still charged for preemptible instances after they are stopped?
After a preemptible instance enters the To Be Released state, does billing continue?
FAQ about preemptible instances without a protection period
FAQ about reserved instances
To what scenarios are regional reserved instances applicable?
How is the instance size flexibility of reserved instances applied?
Do zonal reserved instances provide instance size flexibility?
Can a zonal reserved instance be changed into a regional reserved instance?
Can the scope of a reserved instance be changed from one region to another?
Can I configure a reserved instance to be applied to a specific pay-as-you-go instance?
When does a reserved instance take effect after it is purchased?
After I modify, split, or merge a reserved instance, when does the operation take effect?
Why is the No Upfront payment option not displayed on the buy page?
Can I use Windows reserved instances to offset the image fees of pay-as-you-go Windows instances?
Can I use Linux reserved instances to offset the image fees of pay-as-you-go Linux instances?
Are the consumption details of reserved instances refreshed every hour?
Can a reserved instance be applied to multiple pay-as-you-go instance at the same time?
FAQ about connecting to instances
Workbench
FAQ about Virtual Network Computing (VNC)
Does a VNC management terminal allow multiple users to concurrently log on?
What do I do if a black screen appears while I am connected to a VNC management terminal?
What do I do if a VNC management terminal cannot be accessed?
Why am I unable to use Internet Explorer 8.0 to access a VNC management terminal?
How do I adjust the desktop resolution of a Windows instance?
Why do two cursors appear after I log on to a Windows instance by using a VNC management terminal?
FAQ about using third-party software
FAQ about upgrading and downgrading instance configurations
Can I upgrade the instance types and other configurations of subscription instances?
Can I upgrade the instance types and other configurations of pay-as-you-go instances?
How long does it take to upgrade the configurations of an instance?
How is the fee for an instance configuration upgrade calculated?
Are my cloud service configurations affected if I upgrade the configurations of ECS instances?
I have upgraded the configurations of an instance but no changes have taken effect. Why?
FAQ about managing instances
What do I do if I cannot access a website that runs on an ECS instance?
An ECS instance cannot load the kernel to start. What do I do?
Why am I unable to add sound or video cards to ECS instances?
Can I transfer the unused time of an ECS instance to another ECS instance?
Are public and private IP addresses independent? Can I specify or add IP addresses?
Can load balancing be implemented for a single ECS instance?
Can I change the region of an ECS instance that I purchased?
How do I view subscription ECS instances in all regions in my account?
When can I force stop an ECS instance? What are the consequences?
How do I restore the data that was accidentally deleted from an ECS instance?
How do I handle an ECS instance that remains in the Stopped state?
FAQ about instance security
FAQ about using Linux instances
FAQ about instance limits
FAQ about instance billing
How do I check which instance types are available for purchase in a specific region or zone?
You can go to the Instance Types Available for Each Region page to view the instance types available in each region.
What do I do if resources are sold out when I attempt to purchase an instance in a specific region or zone?
If a specific instance type is sold out when you attempt to purchase an instance of the instance type in a specific region or zone, take one of the following measures:
Select another region.
Select another zone.
Select another instance type.
Instance resources are dynamic. Alibaba Cloud replenishes resources that are insufficient as appropriate. If the instance resources that you want to purchase are still unavailable after you take all of the preceding measures, try again later.
How do I select an instance type that is suitable for my business?
Consider the following factors when you select an instance type. For more information about how to select an instance type, see Best practices for selecting ECS instance types.
Your business requirements
Your website type
The average number of page views per day on your website
The size of your homepage
The data capacity of your website
How long does it take to create an instance?
It takes 1 to 2 minutes to create an ECS instance.
You can connect to a Linux instance immediately after it is created. For more information, see Connect to an ECS instance.
After a Windows instance is created, you must use Sysprep to initialize the operating system before you can connect to the instance. Do not restart the instance while its operating system is being initialized. After the operating system is initialized, you can connect to the instance. For more information, see Connect to an ECS instance. The amount of time that is required to initialize the operating system depends on whether the instance is I/O optimized.
For an I/O optimized Windows instance, it takes 2 to 3 minutes to initialize the operating system.
For a non-I/O optimized Windows instance, it takes 10 minutes to initialize the operating system.
If an error occurs while the instance is being created, submit a ticket.
I paid for an instance but no instance was created. Why?
If resources in the specified zone are insufficient to create an instance of your selected instance type, the instance cannot be created. In this case, a refund is automatically credited to your account. If you do not receive a refund within half an hour, submit a ticket.
You can go to the Instance Types Available for Each Region page to view the instance types available in each region.
Why does a purchased instance have a memory size that is different from the memory size defined in the instance type?
The memory size defined in each instance type is the total memory size, which includes the amount of memory occupied by the operating system, such as Basic Input/Output System (BIOS) reserved memory and memory overheads incurred by the kernel and the hypervisor. As a result, the size of available memory on your instance is smaller than the memory size that is defined in the instance type. Different instance families may use different technology stacks, which causes the percentage of memory occupied by the system to vary slightly.
What are enterprise-level instances? What are entry-level instances?
Enterprise-level instances belong to a series of instance families that Alibaba Cloud introduced in September 2016. Enterprise-level instances provide high performance, consistent computing power, and balanced network performance. Enterprise-level instances have exclusive and consistent computing, storage, and network resources, and are suitable for enterprise scenarios that require high business stability.
Entry-level instances belong to a series of instance families that are intended for individuals or small and medium-sized websites. Entry-level instances share resources, in contrast with enterprise-level instances that each have their own exclusive resources. As a result, entry-level instances do not provide consistent computing performance, but cost less.
What are the differences between enterprise-level and entry-level instances?
Enterprise-level instances use a CPU-bound scheduling scheme. Each vCPU is bound to a CPU hyperthread. Instances do not compete for CPU resources and do provide consistent computing performance as specified in the service-level agreement (SLA).
Entry-level instances use a CPU-unbound scheduling scheme. Each vCPU is randomly allocated to an idle CPU hyperthread. vCPUs of different instances compete for CPU resources, which causes computing performance to fluctuate when traffic loads are heavy. Shared instances can guarantee availability as specified in the SLA, but cannot guarantee performance.
Which instance families are enterprise-level instance families? Which instance families are entry-level instance families?
Among the instance families that are available for purchase, e, t6, t5, s6, n4, mn4, xn4, and e4 are entry-level instance families, and the rest are enterprise-level instance families. For more information, see Overview of instance families.
In what business scenarios do I need to purchase enterprise-level instances?
For information about the business scenarios for which different enterprise-level instances are suitable, see Overview of instance families and Best practices for selecting ECS instance types.
How is the network performance of enterprise-level instances?
The network performance of enterprise-level instances varies based on their specifications. Higher instance specifications indicate higher network performance. For information about the network performance of different instance types, see Overview of instance families.
Which cloud disk categories do enterprise-level instances support?
For information about the cloud disk categories that enterprise-level instances support, see the Disk categories section in the "Disks" topic.
Which image types do enterprise-level instances support?
For information about the public images that enterprise-level instances support, see Overview of public images.
You can also import custom images. For more information, see Import custom images.
What are the limits on changing the instance types of enterprise-level instances?
For information about the limits on changing the instance types of enterprise-level instances, see Instance types and families that support instance type changes.
Can I upgrade an entry-level instance to an enterprise-level instance?
For information about the answer, see Instance types and families that support instance type changes.
What characteristics do persistent memory-optimized instances have?
Persistent memory-optimized instances use high-capacity persistent memory and provide slower access and higher data durability than instances that use regular memory. When persistent memory-optimized instances are stopped or restarted, data in their persistent memory is not lost. Persistent memory can be used as memory or local disks.
When persistent memory is used as memory, you can move some data from regular memory to persistent memory, such as non-hot data that does not require high-speed storage access. Persistent memory offers large capacity at a low per-GiB price and can help reduce the total cost of ownership (TCO) per GiB of memory.
When persistent memory is used as local disks, it functions like local SSDs to deliver ultra-high performance and a read/write latency as low as 400 nanoseconds. In this case, you can use persistent memory for core application databases that require consistent response time. You can also replace cache disks with persistent memory to achieve higher IOPS, higher bandwidth, and lower latency and improve cluster-wide business performance.
What requirements do persistent memory-optimized instances have for operating systems?
The following image versions can be used on persistent memory-optimized instances:
Alibaba Cloud Linux 2
CentOS 7.6 and later
Ubuntu 18.04 and 20.04
Alibaba Cloud provides support for Alibaba Cloud Linux 2. Alibaba Cloud Linux 2 integrates ready-to-use tools in scenarios for which persistent memory is suitable. In some scenarios, such as those in which Redis applications are used, Alibaba Cloud Linux 2 outperforms community-supported Linux distributions by 20%.
After I purchase a persistent memory-optimized instance, how do I configure the persistent memory of the instance as memory?
You can use tools to configure the persistent memory as memory. For more information, see Configure the usage mode of persistent memory.
When persistent memory is being used as memory on persistent memory-optimized instances, can I deploy and run Redis applications on the instances?
You can significantly reduce the TCO per GiB of memory for Redis applications by running them on persistent memory-optimized instances. To ensure performance, modify the Redis applications by stratifying their data and storing the data in different types of memory. You can store non-hot data in persistent memory and hot data in regular memory.
To reduce your modification costs, Alibaba Cloud provides re6p instance types exclusively for Redis applications. To deploy Redis applications on re6p instances, you need to only run several commands. For more information, see Deploy Redis on persistent memory-optimized instances.
When you purchase persistent memory-optimized instances, select instance types that are named in the ecs.re6p-redis.<nx>large format.
I have a Redis cluster that is based on regular memory. How do I migrate business from this cluster to persistent memory-optimized instances? What items do I need to take note of?
Ensure business consistency and data reliability while you are migrating business from the Redis cluster to persistent memory-optimized instances. Before the migration, we recommend that you purchase a persistent memory-optimized instance and test it with a small amount of required business traffic to check whether the basic performance and capacity model of the instance meet your business requirements. If the basic performance and capacity model of the instance meet your business requirements, you can scale out to more persistent memory-optimized instances to take over all business from the entire Redis cluster.
Can I deploy and run a PS on persistent memory-optimized instances whose persistent memory is used as memory?
Traditionally, server nodes in a PS store all training parameters of training clusters in their memory. These parameters consume a large amount of memory and are expensive to store. When you run a PS on persistent memory-optimized instances whose persistent memory is used as memory, we recommend that you store all parameters in persistent memory and only hash tables in regular memory. This way, you can significantly reduce the TCO of training clusters.
You can modify the applications based on your business requirements, or submit a ticket to contact Alibaba Cloud technical support.
After I purchase a persistent memory-optimized instance, how do I configure the persistent memory of the instance as local disks?
You can use tools to configure the persistent memory as local disks. For more information, see Configure the usage mode of persistent memory.
What applications can use local disks that can deliver higher performance?
To optimize performance or costs for I/O-intensive applications, you can select persistent memory-optimized instances. Persistent memory-optimized instances can address common issues such as the following ones:
The latency of individual SQL queries is high, and SQL queries require a more consistent response time.
It takes longer than required to load resources on frontend game servers, heavily loaded databases, and heavily loaded web servers.
You purchased large cloud disks or local disks for their high IOPS and bandwidth, but have unused capacity that causes unnecessary costs.
You can use persistent memory-optimized instances in typical I/O-intensive scenarios such as the following ones:
Redis and other NoSQL databases such as Cassandra and MongoDB
Structured databases such as MySQL
I/O-intensive applications such as e-commerce, online games, and media applications
Search scenarios that use solutions such as Elasticsearch
Live video streaming, instant messaging, and room-based online games that require persistent connections
High-performance relational databases and OLTP systems
Can I deploy Redis or MySQL applications on persistent memory-optimized instances whose persistent memory is used as local disks? Do the applications need to be modified in the same manner as they do when persistent memory is used as memory?
Yes, you can deploy Redis and MySQL applications on persistent memory-optimized instances whose persistent memory is used as local disks. You do not need to modify these applications before they can recognize the persistent memory as standard SSDs.
How is the performance of persistent memory used as local disks, compared with local NVMe SSDs and cloud disks?
The following table describes the performance comparison between local NVMe SSDs, enhanced SSDs (ESSDs), and persistent memory-based local disks.
The performance data in the following table is only for reference. Data in the results of your tests prevails.
Metric | 128-GiB persistent memory | 1,788-GiB NVMe SSD | 800-GiB ESSD at performance level 1 (PL1) |
Read bandwidth | 8 GB/s to 10 GB/s | 2 GB/s to 3 GB/s | 0.2 GB/s to 0.3 GB/s |
Read/write bandwidth | 8 GB/s to 10 GB/s | 1 GB/s to 2 GB/s | 0.2 GB/s to 0.3 GB/s |
Write bandwidth | 2 GB/s to 3 GB/s | 1 GB/s to 2 GB/s | 0.2 GB/s to 0.3 GB/s |
Read IOPS | 1,000,000 | 500,000 | 20,000 to 30,000 |
Read/write IOPS | 1,000,000 | 300,000 | 20,000 to 30,000 |
Write IOPS | 1,000,000 | 300,000 | 20,000 to 30,000 |
Read latency | 300 nanoseconds to 400 nanoseconds | 100,000 nanoseconds | 250,000 nanoseconds |
Write latency | 300 nanoseconds to 400 nanoseconds | 20,000 nanoseconds | 150,000 nanoseconds |
You can perform the following steps to test the performance of the persistent memory when it is used as local disks:
Configure persistent memory as local disks and attach the disks to instances.
For more information, see Configure the usage mode of persistent memory.
Use disk performance test tools to test the performance of the persistent memory that is used as local disks.
For information about how to use fio to test disk performance in Linux, see the "Commands used to test the performance of local disks" section in Test the performance of block storage devices.
How reliable is persistent memory?
The reliability of data stored in persistent memory varies based on the reliability of persistent memory devices and the physical servers to which these devices are attached. Risks of single points of failure exist. To ensure the reliability of application data, we recommend that you implement data redundancy at the application layer and use cloud disks for long-term data storage.
When persistent memory-optimized instances are released, data stored in their persistent memory is automatically cleared. Before you release the instances, we recommend that you back up the data. It takes longer to release persistent memory-optimized instances than to release other types of instances.
What users and scenarios are u1 instances suitable for?
u1 instances are intended for small and medium-sized enterprise users and are suitable for most general-use scenarios that do not require high CPU capacity or high performance, such as web applications and websites, enterprise office applications, offline data analytics, and small and medium-sized databases.
What scenarios are u1 instances unsuitable for?
For scenarios that require high computing capacity and high performance to provide optimal user experience, such as gaming and high-frequency trading, we recommend that you use c7, g7, and r7 instances.
u1 instances can be deployed to different server platforms. Business performance may significantly vary between server platforms. For scenarios that have strict requirements for business performance consistency, such as high-availability scenarios in which the performance of primary nodes must be consistent with the performance of secondary nodes, we recommend that you use c7, g7, and r7 instances.
What processor models do u1 instances support?
u1 instances support Intel ® Xeon® Platinum 6149 (Skylake), 8163 (Skylake), 8269CY (Cascade Lake), 8260 (Cascade Lake), 8369HB (Cooper Lake), and 8369HC (Cooper Lake) processors.
Can I specify processor models for u1 instances?
No, you cannot specify processor models for u1 instances.
How many data disks can be attached to a u1 instance?
The maximum number of data disks that can be attached to a u1 instance varies based on the instance type.
Up to eight data disks can be attached to an ecs.u1-c1m8.xlarge or smaller instance.
Up to 16 data disks can be attached to an ecs.u1-c1m1.2xlarge or larger instance.
At what base frequency and turbo frequency do the processors of u1 instances run?
u1 instances can be deployed to server platforms that provide different levels of computing power. The CPU base frequency and turbo frequency of u1 instances vary based on the CPU models of the platforms where the instances reside. In addition, u1 instances may be migrated across CPU platforms during the instance lifecycle. This way, the CPU base frequency and turbo frequency of the instances may change accordingly.
Why may u1 instances be migrated across server platforms during the instance lifecycle?
During the instance lifecycle, u1 instances may be migrated across hosts due to reasons such as the upgrade or maintenance of the underlying host. The original and new hosts may use different processors. All types of instances may be migrated across hosts during the instance lifecycle. For information about the differences between pre-migration and post-migration processors of u1 instances, see Universal instance families.
What scenarios are e instances unsuitable for?
e instances are unsuitable for enterprise-level scenarios that require consistent performance and a performance SLA.
What CPU models do e instances support? Can I specify processor models for e instances?
e instances support Intel® Xeon® Platinum Scalable processors. You cannot specify processor models for e instances.
e instances are built on mainstream, cost-effective CPU platforms and may be migrated across CPU platforms during the instance lifecycle. e instances use technological capabilities to promote cross-platform compatibility.
At what base frequency and turbo frequency do the processors of e instances run?
The processors of e instances run at a base frequency of 2.5 GHz, and no commitments regarding turbo boost are made because the instances may be migrated across CPU platforms during the instance lifecycle.
Why may e instances be migrated across server platforms during the instance lifecycle?
During the instance lifecycle, e instances may be migrated across hosts due to reasons such as the upgrade or maintenance of the underlying host. The original and new hosts may use different processors.
You do not need to take note of underlying hosts. e instances are designed based on the resource pooling technology and intelligent scheduling algorithm of Alibaba Cloud to allow dynamic resource management, provide consistent computing capacity, stability, and scalability, and guarantee the provision of resources for your applications.
What are the differences between ECS bare metal instances, traditional cloud hosts (virtual machines), and traditional physical machines?
For information about the differences, see Overview of ECS Bare Metal Instance families.
How is the network performance of ECS bare metal instances?
The network performance of ECS bare metal instances varies based on their specifications. Higher instance specifications indicate higher network performance. For information about the network performance of different instance types, see Overview of instance families.
Which cloud disk categories are supported by ECS bare metal instances? How many data disks can be attached to an ECS bare metal instance?
ECS bare metal instances support ultra disks, standard SSDs, and Enterprise SSDs (ESSDs). Up to 16 data disks can be attached to an ECS bare metal instance.
Can the configurations of ECS bare metal instances be upgraded or downgraded? Do the instances support failover?
No, the configurations of ECS bare metal instances cannot be upgraded or downgraded. The instances do support failover. When the physical machine that hosts an ECS bare metal instance fails, the instance fails over to a healthy physical machine. Data is retained in the data disks of the instance.
How do I create an SCC instance?
You can create an SCC instance by using one of the following methods:
If you want to use only RDMA, create an SCC instance in the ECS console. For more information, see Create an SCC instance.
If you want to use RDMA, the High Performance Computing (HPC) scheduler, and the cluster resizing feature, create an SCC and then an SCC instance in the E-HPC console.
How are SCC instances billed?
SCC instances can be billed as a weekly, monthly, or yearly subscription.
How do I create an SCC by using E-HPC?
You can log on to the E-HPC console or call the CreateCluster operation to create an SCC.
How do I use SCC RDMA?
When you create an SCC instance, select an image that is customized for SCC. The image supports RDMA over Converged Ethernet (RoCE) drivers and OpenFabrics Enterprise Distribution (OFED) stacks. Then, you can use SCC RDMA by virtue of the InfiniBand (IB) verbs API or implement RDMA-based communication by using the Message Passing Interface (MPI).
Which instance types support preemptible instances?
Instance types that support pay-as-you-go instances support preemptible instances. If a preemptible instance of a specific instance type cannot be created due to insufficient resources, try another instance type.
In which regions can I create preemptible instances?
Preemptible instances can be created in all Alibaba Cloud regions. If preemptible instances cannot be created in a specific region due to insufficient resources, try another region.
When I attempt to purchase an ECS instance, the Preemptible Instance option is unavailable on the instance buy page. Why?
The availability of the Preemptible Instance option depends on your ECS usage. The Preemptible Instance option is available only to authorized users.
How do I bid for a preemptible instance?
To create a preemptible instance, you must specify a maximum hourly price to bid for spare resources of an instance type. If the spot price (current market price) of the instance type does not exceed your specified maximum hourly price, the preemptible instance is created and billed based on the spot price. For more information, see Create a preemptible instance.
What is the relationship between the user-defined maximum hourly price and the spot price of a preemptible instance?
If the spot price does not exceed the user-defined maximum hourly price, the preemptible instance is created and billed based on the spot price. After a preemptible instance is created, it enters a protection period during which it cannot be automatically released due to insufficient resources or fluctuations in the spot price.
After the protection period ends, the system checks the spot price and resource availability of the instance type every 5 minutes. If the spot price exceeds the user-defined maximum hourly price or if resources for the instance type are insufficient, the preemptible instance that is in the Running state is released.
Can I view the spot price of an instance type when I purchase a preemptible instance?
Yes, when you create a preemptible instance in the ECS console, you can view the spot price and price history of each selected instance type. The total instance price includes fees for the instance type, storage, and bandwidth and is displayed in the lower part of the instance buy page. The instance type fee is the spot price of the selected instance type.
Can I view the price history of a preemptible instance type? How?
Yes, you can select an instance type to view its price history when you attempt to create a preemptible instance in the ECS console. You can also call the DescribeSpotPriceHistory operation to view the price history of a preemptible instance type.
How many preemptible instances can an account purchase?
vCPU-based quotas, not instance quotas, limit the number of preemptible instances in each account. When you create a preemptible instance, you can view the available vCPU-based quota after you select an instance type. For more information, see the "Instance limits" section in Limits.
How do I increase my vCPU-based quotas?
To increase your vCPU-based quotas, submit a ticket.
Can preemptible instances be converted into subscription instances?
No, preemptible instances cannot be converted into subscription instances.
Can the instance type of a preemptible instance be changed?
No, the instance type of a preemptible instance cannot be changed.
A preemptible instance does not meet my requirements. How do I upgrade it to a higher-specification instance type at a low cost?
The instance types of preemptible instances cannot be changed. We recommend that you create a custom image from the system disk of your preemptible instance and then use the custom image to create a preemptible instance of a higher-specification instance type. For more information, see Create a custom image from a snapshot and Create an ECS instance by using a custom image.
If I select Use Automatic Bid (SpotAsPriceGo) when I create a preemptible instance, will the created instance be automatically released due to insufficient resources?
When you select Use Automatic Bid (SpotAsPriceGo), the spot price is used as the bid price and can rise as high as the pay-as-you-go price for the selected instance type. This ensures that the created preemptible instance will not be automatically released due to fluctuations in the spot price. However, the instance may still be automatically released due to insufficient resources for the instance type.
If I set the maximum hourly price to the pay-as-you-go price for the selected instance type when I create a preemptible instance, may the instance be automatically released?
You can set the maximum hourly price to the pay-as-you-go price for your selected instance when you create a preemptible instance. This ensures that the created instance will not be automatically released due to fluctuations in the spot price. However, the instance may still be automatically released due to insufficient resources for the instance type.
Can I cancel or reschedule the automatic release of a preemptible instance?
Yes, you can cancel or reschedule the automatic release of a preemptible instance at any time before the instance is released.
I do not have overdue payments in my account. Why have my preemptible instances been released?
After the protection period of a preemptible instance ends, the instance is automatically released if the spot price exceeds your bid price or if resources are insufficient.
How are release rates on the instance buy page calculated? Why are most release rates in the range of 0% to 3%?
Release rates vary based on zones and instance types. The single-day release rate of an instance type in a zone is calculated by using the following formula: Number of released preemptible instances on a single day/Total number of preemptible instances on a single day. Most release rates are in the range of 0% to 3% as expected. Alibaba Cloud continues efforts to reduce release rates and ensure higher availability of preemptible instances.
Am I notified when my preemptible instances are about to be released? How am I notified?
Yes, you are notified when your preemptible instances are about to be released. When your preemptible instance needs to be released due to a spot price change or insufficient resources, the instance first enters the To Be Released state and is automatically released in 5 minutes.
You can use CloudMonitor to subscribe to notifications for preemptible instance interruption events. For more information, see Subscribe to ECS system event notifications.
To check whether an instance is in the To Be Released state, you can view the instance metadata, or call the DescribeInstances operation and view the returned OperationLocks data. For more information, see Overview of ECS instance metadata and DescribeInstances.
Can the data of a preemptible instance be automatically retained when the instance is released?
No, the data of a preemptible instance cannot be automatically retained when the instance is released. When a preemptible instance is no longer needed, we recommend that you create snapshots to back up your data and environment and then release the instance. You can purchase preemptible instances at any time. For more information, see Create a snapshot.
What do I do if I cannot restart a preemptible instance that was stopped in economical mode?
An instance that was stopped in economical mode may be unable to start due to insufficient resources. To save the business data that is stored on the instance, perform the following steps:
Create a snapshot for each disk on the instance (Instance A). For more information, see Create a snapshot.
If the creation progress of a snapshot is 100%, the snapshot is created.
Use the created snapshots to create a custom image. For more information, see Create a custom image from a snapshot.
If the image is in the Available state, the image is created.
Create an instance (Instance B) from the custom image.
For more information, see Create an ECS instance by using a custom image.
Check whether the business data on Instance B is correct.
If the business data is correct, release instance A. For more information, see Release an instance.
If the business data is incorrect, detach disks from Instance A, attach the disks to Instance B, and then release Instance A. When you detach disks from Instance A, clear Release Disk with Instance. For more information, see Detach a data disk and Release an instance.
When the protection period of a preemptible instance ends, the preemptible instance enters the To Be Released state if the spot price exceeds the bid price. During the 5-minute period in which the preemptible instance is pending release, can the preemptible instance be retained if the bid price is higher than the spot price?
No, the preemptible instance cannot be retained. Before a preemptible instance is released, you are notified of the release and the instance enters the To Be Released state. The status of the instance cannot be changed and the instance is automatically released after approximately 5 minutes.
How do I retain the public IP address of a preemptible instance when the instance is released?
When a preemptible instance is released, the public IP address of the instance is reclaimed. If you want to retain the public IP address of the instance, we recommend that you use Elastic IP Address (EIP). For more information, see Elastic IP addresses.
You can convert the static public IP address to an EIP. For more information, see Convert the public IP address of an ECS instance in a VPC to an EIP and Convert the system-assigned public IP address of an instance in the classic network into an EIP.
How do I view the bid price of a preemptible instance?
Go to the Instance Details page of a preemptible instance and view the bid price of the instance in the Billing Information section.
How is a preemptible instance billed if its lifespan is shorter than 1 hour?
If a preemptible instance has a lifespan that is shorter than 1 hour, the instance is billed based on its lifespan in seconds. The hourly price of the instance is the spot price that was in effect when the instance was created. The per-second price of the instance is calculated by using the following formula: Per-second price = Hourly price/3,600. The fees for a preemptible instance from creation to release are accurate to two decimal places. Accrued costs of less than USD 0.01 for a preemptible instance are not charged.
To what resources are the prices of preemptible instances applicable?
The prices of preemptible instances are applicable only to instance types. You are charged for other resources such as system disks, data disks, and network bandwidth at the same prices as those of pay-as-you-go instances.
Why do preemptible instances of the same instance type and region vary widely in spot prices across different zones?
Price differences are caused by resource stock differences. Each zone corresponds to different data centers. In each region, data centers for zones may vary in terms of construction conditions, capacity, sales, and deployment policies. The inventory of each instance type may also vary across different zones. The spot prices of preemptible instances fluctuate with changes in supply and demand and are therefore different across zones.
May the spot price of a preemptible instance exceed the pay-as-you-go price?
No, the spot price of a preemptible instance can be up to but cannot exceed the pay-as-you-go price. You do not need to worry about being charged more for a preemptible instance than for a pay-as-you-go instance with the same configurations.
For the first hour after a preemptible instance is created, am I charged at a price that fluctuates with the spot price?
No, the hourly price of the first instance hour for a preemptible instance is set at the beginning of the hour and remains in effect for the whole hour.
Am I still charged for preemptible instances after they are stopped?
Yes, you are still charged for your preemptible instances after they are stopped. When a preemptible instance is no longer needed, we recommend that you create snapshots to back up your data and environment and then release the instance. You can purchase preemptible instances at any time.
You are still charged for preemptible instances after you stop them by using the ECS console or by calling the StopInstance operation.
I purchased a preemptible instance at 08:00 with a bid of USD 2 per hour, and the spot price of the preemptible instance was USD 2 per hour at 08:30. Why was I charged USD 1.5 per hour for the instance from 08:30 to 09:00?
The following figure shows the fluctuations of the spot price.
Fee for the protection period of 8:00:00 to 9:00:00 (the transaction price is USD 1.5 per hour and the usage duration is 60 minutes): (1.5/3,600) × 60 × 60 = USD 1.5.
Fee for the period from 9:00:00 to 9:30:00 (the spot price is USD 0.5 per hour and the usage duration is 30 minutes): (0.5/3,600) × 30 × 60 = USD 0.25.
Fee for the period from 9:30:00 to 10:00:00 (the spot price is USD 1 per hour and the usage duration is 30 minutes): (1/3,600) × 30 × 60 = USD 0.5.
From 08:30:00 to 09:00:00, the preemptible instance was in the protection period. You were charged for the instance based on the spot price at the time of purchase, which was USD 1.5 per hour. After the protection period ends, you are charged for the instance based on the real-time spot price.
After a preemptible instance enters the To Be Released state, does billing continue?
After a preemptible instance enters the To Be Released state, it is automatically released in 5 minutes. During the 5-minute period, the instance continues to run and be billed. Billing stops only when the instance is released.
Which is more cost-effective: a preemptible instance with a protection period or a preemptible instance without a protection period?
Preemptible instances without a protection period are more cost-effective and cost about 10% less than preemptible instances with a protection period.
Is the release rate of preemptible instances without a protection period higher than that of preemptible instances with a protection period?
You can check the release rate of each instance type on the instance buy page in the ECS console, regardless of whether you attempt to create preemptible instances with or without a protection period. For each instance type, the release rate is determined by the bidding policy and the supply and demand of resources.
Are preemptible instances without a protection period prioritized for release over those with a protection period?
No, preemptible instances without a protection period are not prioritized for release over those with a protection period.
If I have created a preemptible instance without a protection period, do subsequently created preemptible instances also not have a protection period?
No, preemptible instances are created without a protection period only if you choose not to configure a protection period for them. By default, each preemptible instance is created with a protection period of 1 hour.
You can configure a protection period for preemptible instances only when you create the instances.
If a preemptible instance does not have a protection period, am I still notified 5 minutes before the instance is released?
Yes, you are still notified 5 minutes before a preemptible instance is released even if the instance does not have a protection period.
Are fewer resources provided for preemptible instances without a protection period than for those with a protection period?
No, the same number of resources are provided for preemptible instances with and without a protection period.
Can I convert between preemptible instances with and without a protection period?
No, you cannot convert between preemptible instances with and without a protection period. By default, each preemptible instance is created with a protection period of 1 hour. You can configure a protection period for preemptible instances only when you create them. The protection period settings of preemptible instances cannot be changed after the instances are created.
What are reserved instances?
A reserved instance is a discount coupon that can be automatically applied to one or more pay-as-you-go instances (excluding preemptible instances) to offer billing discounts. Reserved instances can also be used to reserve instance resources. A combination of reserved instances and pay-as-you-go instances provides higher cost-effectiveness and a higher degree of flexibility than subscription instances.
Do reserved instances provide reserved resources?
Zonal reserved instances provide reserved resources, but regional reserved instances do not.
What operating systems do reserved instances support?
Reserved instances support Windows and Linux. For example, you can purchase a reserved Linux instance and then apply this reserved instance to pay-as-you-go Linux instances that match the attributes of the reserved instance regardless of the image type, such as public images, custom images, shared images, or Alibaba Cloud Marketplace images.
To apply a reserved instance to pay-as-you-go instances that were created from Bring Your Own License (BYOL) images, submit a ticket.
Which instance families do reserved instances support?
Reserved instances support the following instance families:
- General-purpose instance families: g7, g7a, g6e, g6, g5, g5ne, and sn2ne
- Compute-optimized instance families: c7, c7a, c6e, c6, c5, ic5, and sn1ne
- Memory-optimized instance families: r7, r7a, r6e, r6, r5, re6, re4, and se1ne
- Big data instance family: d2s
- Instance families with local SSDs: i3, i3g, i2, i2g, and i2gne
- Instance families with high clock speeds: hfg7, hfc7, hfr7, hfg6, hfc6, hfr6, hfg5, and hfc5
- GPU-accelerated compute-optimized instance families: gn7, gn6i, gn6e, gn6v, gn5, and gn5i
- ECS Bare Metal Instance families: ebmgn7, ebmgn6i, ebmgn6e, ebmg6, ebmc6, ebmr6, ebmhfg6, ebmhfc6, and ebmhfr6
- Burstable instance families: t6 and t5
Can reserved instances be applied to preemptible instances?
No, reserved instances cannot be applied to preemptible instances.
Can the instance families of reserved instances be changed?
No, the instance families of reserved instances cannot be changed.
To what scenarios are zonal reserved instances applicable?
We recommend that you purchase zonal reserved instances when you have clear requirements to reserve resources.
To what scenarios are regional reserved instances applicable?
We recommend that you purchase regional reserved instances if you want to have a higher degree of zone flexibility or instance size flexibility.
How is the zone flexibility of reserved instances applied?
Only regional reserved instances provide zone flexibility. Example:
Assume that you are running the following pay-as-you-go instance:
One ecs.c5.xlarge Linux instance in Qingdao Zone B. This instance is named C5PAYG-b.
Assume that you purchased the following reserved instance:
One regional ecs.c5.xlarge reserved instance in the China (Qingdao) region. This instance is named C5RI.
C5RI is matched to C5PAYG-b.
You release C5PAYG-b and create another pay-as-you-go Linux instance of the same instance type named C5PAYG-c in Qingdao Zone C. C5RI is matched to C5PAYG-c.
How is the instance size flexibility of reserved instances applied?
Only regional reserved instances provide instance size flexibility. Examples:
Assume that you have one regional ecs.g5.4xlarge reserved instance. The reserved instance can be applied to one ecs.g5.4xlarge pay-as-you-go instance, two ecs.g5.2xlarge pay-as-you-go instances, or four ecs.g5.xlarge pay-as-you-go instances.
Assume that you have a one-year regional ecs.g5.xlarge reserved instance. The reserved instance can be applied to offset the hourly fees of an ecs.g5.xlarge pay-as-you-go instance or 50% of the hourly fees of an ecs.g5.2xlarge pay-as-you-go instance for one year.
Do zonal reserved instances provide instance size flexibility?
No, zonal reserved instances do not provide instance size flexibility. A zonal reserved instance can be applied only to pay-as-you-go instances of the same instance type.
Do zonal reserved instances provide zone flexibility?
No, zonal reserved instances do not provide zone flexibility. A zonal reserved instance can be applied only to pay-as-you-go instances in the same zone as it.
Can a zonal reserved instance be changed into a regional reserved instance?
Yes, a zonal reserved instance can be changed into a regional reserved instance. The scope of a reserved instance can be changed in the following ways:
From a zone to a region
From a region to a zone
From one zone to another in the same region
Can the scope of a reserved instance be changed from one region to another?
No, the scope of a reserved instance cannot be changed from one region to another. For example, assume that you have a zonal reserved instance scoped to Hangzhou Zone B. You can change the instance scope to another zone in the China (Hangzhou) region or change the instance into a regional reserved instance scoped to the China (Hangzhou) region. However, you cannot change the scope of the instance to a zone in another region or change the instance into a regional reserved instance scoped to another region.
Can reserved instances be used across accounts?
No, reserved instances cannot be used across accounts.
Can reserved instances be used to offset the storage and network usage fees of pay-as-you-go instances?
No, reserved instances cannot be used to offset the storage or network usage fees of pay-as-you-go instances. Reserved instances are applied to offset the fees for vCPUs and memory of pay-as-you-go instances. Windows reserved instances can also offset image fees.
Can I configure a reserved instance to be applied to a specific pay-as-you-go instance?
No, you cannot configure a reserved instance to be applied to a specific pay-as-you-go instance. When multiple pay-as-you-go instances match the attributes of a reserved instance, the reserved instance is applied based on the optimized matching scheme.
How are reserved instances billed?
Reserved instances are billed separately and support the All Upfront, Partial Upfront, and No Upfront payment options.
The term of a reserved instance begins at the time when the instance is purchased. You are charged based on the payment option that you selected, regardless of whether the reserved instance is matched to pay-as-you-go instances. The All Upfront option is most cost-effective.
When does a reserved instance take effect after it is purchased?
A reserved instance takes effect and is billed starting on the hour when the reserved instance is purchased. The reserved instance expires at 00:00:00 on the day after its term end date. Examples:
Assume that you purchase a reserved instance with a one-year term at 13:45:00 on February 26, 2019. The reserved instance takes effect, starts to be billed at 13:00:00 on February 26, 2019, and expires at 00:00:00 on February 27, 2020. If you already have eligible pay-as-you-go instances when you purchase a reserved instance, the reserved instance is applied to offset the fees generated by the pay-as-you-go instances starting from the hour of 13:00:00 to 14:00:00 on February 26, 2019 until the reserved instance expires.
Assume that you purchase a reserved instance with a one-year term at 13:45:00 on February 26, 2019. The reserved instance takes effect, starts to be billed at 00:00:00 on March 1, 2019, and expires at 00:00:00 on March 1, 2020. If you already have eligible pay-as-you-go instances when you purchase a reserved instance, the reserved instance is applied to offset the fees generated by the pay-as-you-go instances starting from 00:00:00 on March 1, 2019 until the reserved instance expires.
After I modify, split, or merge a reserved instance, when does the operation take effect?
When reserved instances are modified, split, or merged, new reserved instances are generated and the original ones become invalid. The new reserved instances take effect and the original ones become invalid on the hours of the modification, splitting, or merging operations. For example, assume that you split one ecs.g5.2xlarge zonal reserved instance RI1 into two zonal ecs.g5.xlarge reserved instances RI2 and RI3 at 13:45:00 on February 26, 2019. At 13:00:00 on February 26, 2019, RI1 becomes invalid and RI2 and RI3 take effect. Starting from 13:00:00 on February 26, 2019, the instance type eligible for resource reservation and billing discounts is ecs.g5.xlarge, no longer ecs.g5.2xlarge. If RI2 and RI3 are matched to pay-as-you-go instances immediately after they take effect, RI2 and RI3 are applied to offset the hourly fees of ecs.g5.xlarge pay-as-you-go instances starting from 13:00:00 on February 26, 2019.
Why is the No Upfront payment option not displayed on the buy page?
The availability of this option depends on your ECS usage.
Can the payment option of a reserved instance be changed?
No, the payment options of reserved instances cannot be changed.
Can reserved instances be resold?
No, reserved instances cannot be resold.
Can I use Windows reserved instances to offset the image fees of pay-as-you-go Windows instances?
Yes, Windows reserved instances can be applied to offset the image fees of pay-as-you-go Windows instances. This is because Windows reserved instances already include Windows images at no additional cost.
Can I use Linux reserved instances to offset the image fees of pay-as-you-go Linux instances?
No, Linux reserved instances cannot be applied to offset the image fees of pay-as-you-go Linux instances.
Are the consumption details of reserved instances refreshed every hour?
Yes, the consumption details of reserved instances are refreshed every hour.
Can a reserved instance be applied to multiple pay-as-you-go instance at the same time?
Yes, a reserved instance can be applied to more than one pay-as-you-go instance at the same time. The reserved instance checks for eligible pay-as-you-go bills on an hourly basis and deducts fees based on its computing power.
The computing power and term of each reserved instance are fixed. You cannot increase the computing power of a reserved instance by shortening its term.
For example, assume that you have a reserved instance with the following attributes:
Instance type: c5.large
Instances: 1 (indicating that the reserved instance can match one pay-as-you-go instance of the specified instance type)
Term: one year
The following examples demonstrate how the reserved instance is applied based on the pay-as-you-go instances that exist:
Assume that six c5.large pay-as-you-go instances exist for 1 hour each. Each of these pay-as-you-go instances consumes 1 hour of computing power equal to that delivered by the c5.large reserved instance every hour. The reserved instance is randomly applied to one of these pay-as-you-go instances. You cannot configure the reserved instance to be applied to all six pay-as-you-go instances by shortening the term of the reserved instance to two months.
Assume that six c5.large pay-as-you-go instances exist for 10 minutes each. The six instances consume 10 minutes of computing power each and in total consume the amount of computing power that the c5.large reserved instance can deliver every hour. The reserved instance is applied to all six pay-as-you-go instances.
Assume that six c5.large pay-as-you-go instances exist for 15 minutes each. The six instances consume 15 minutes of computing power each, in total exceeding the amount of computing power that the c5.large reserved instance can deliver every hour. The reserved instance is randomly applied to the pay-as-you-go instances to offset the fees for 1 hour of computing power.
When I attempted to connect to a Linux instance by using Workbench, the connection failed and I was prompted that the connection timed out. Why?
This may be caused by one of the following issues:
The instance is not in the Running state.
No rules in the security groups of the instance allow traffic on the specified port. For more information about security group rules, see Connect to a Linux instance by using a password or key.
When I attempted to connect to a Linux instance by using Workbench, the connection failed and I was prompted that the connection was denied. Why?
This may be caused by one of the following issues:
The remote connection service that corresponds to SSH or Remote Desktop Protocol (RDP) is not enabled on the instance.
Ports used to accept remote connections, such as SSH port 22 and RDP port 3389, are not open on the instance.
When I attempted to connect to a Linux instance by using Workbench, the connection failed and I was prompted that the username or password was incorrect. Why?
This is because the username or password that you entered is incorrect. Perform the following operations:
Enter the correct username. The username can be root, ecs-user, or a username that you specified when you created the instance.
Enter the correct password. If you forget your password, you can reset it. For more information, see Reset the logon password of an instance.
To use the root username to log on to the Linux instance, ensure that the /etc/ssh/sshd_config file contains the
PermitRootLogin yes
setting. For more information, see Connect to a Linux instance by using a password or key.
When I attempted to connect to a Windows instance by using Workbench, the connection failed and I was prompted that the connection timed out. Why?
This may be caused by one of the following issues:
The instance is not in the Running state.
No rules in the security groups of the instance allow traffic on the specified port. For more information about security group rules, see Connect to a Linux instance by using a password or key.
The remote connection service that corresponds to SSH or RDP is not enabled on the instance.
Ports used to accept remote connections, such as SSH port 22 and RDP port 3389, are not open on the instance.
When I attempted to connect to a Windows instance by using Workbench, the connection failed and I was prompted that the username or password was incorrect. Why?
This is because the username or password that you entered is incorrect. Perform the following operations:
Enter the correct username. In most cases, the username of Windows instances is administrator.
Enter the correct password. If you forget your password, you can reset it. For more information, see Reset the logon password of an instance.
If you log on to the Windows instance as a non-administrator user, the user must belong to the Remote Desktop Users group.
When I attempted to connect to a Windows instance, the error code 0x112f was returned. Why?
If the 0x112f
error code is returned, the Windows instance may have insufficient memory. For information about how to resolve this issue, see What do I do if I am prompted with the 0x112f error code when I connect to a Windows instance over RDP?
Can I decompress files by using Workbench?
No, you cannot use Workbench to decompress files.
In Linux, you can compress files in various formats. You can run commands to install decompression tools and decompress files. The following table describes the commands.
To view information about the parameters used in the decompression tools, run the <Decompression tool name> --help
command. Example commands: tar --help
, gzip --help
, bzip2 --help
, and unzip --help
.
File format | Decompression tool | Command used to decompress files |
.tar |
|
|
.tar.gz |
| |
.tar.bz2 |
| |
.tar.xz |
| |
.gz |
|
|
.bz2 |
|
|
.zip |
|
|
Does a VNC management terminal allow multiple users to concurrently log on?
No, a VNC management terminal allows only one user to log on at a time.
What do I do if a black screen appears while I am connected to a VNC management terminal?
A black screen indicates that the instance is in sleep mode. Perform the following operations based on the instance operating system:
For a Linux instance, click your mouse or press any key to activate the instance.
For a Windows instance, choose
in the upper-left corner to go to the logon page.
What do I do if a VNC management terminal cannot be accessed?
You can use a browser to access the VNC management terminal for troubleshooting. For example, you can use Google Chrome to access the VNC management terminal and press the F12 key to open the Developer tools panel. Then, click the Console tab and identify errors based on the information displayed.
Why am I unable to use Internet Explorer 8.0 to access a VNC management terminal?
VNC management terminals support Internet Explorer 10 and later.
We recommend that you use Google Chrome because it is most compatible with the Alibaba Cloud Management Console.
When I use Firefox to access a VNC management terminal, an error message is returned indicating that a secure connection cannot be established. What do I do?
This problem occurs if the encryption algorithm used by your Firefox version is different from the encryption algorithm used by the VNC management terminal.
We recommend that you use Google Chrome because it is most compatible with the Alibaba Cloud Management Console.
How do I connect to a Linux instance?
Linux instances support SSH connection. You can use one of the following methods to connect to a Linux instance:
What are the default username and password used to connect to the operating system of an ECS instance?
The default username varies based on the operating system type.
For Windows instances, the default username is
administrator
.For Linux instances, the default username is
root
orecs-user
.
The password used to connect to the operating system of an instance is set by you during instance creation. For more information, see Create an instance on the Custom Launch tab. If you forget the password, you can reset it. For more information, see Reset the logon password of an instance.
How do I adjust the desktop resolution of a Windows instance?
You can use one of the following methods to adjust the desktop resolution of a Windows instance:
Connect to the instance by using a VNC management terminal and adjust the desktop resolution of the instance.
In this example, an instance that runs a
Windows Server 2019 Datacenter 64-bit
operating system is used. Perform the following operations to adjust the desktop resolution of the instance:Connect to the instance by using a VNC management terminal in the ECS console. For more information, see Connect to a Windows instance by using a password.
On the Windows desktop, right-click a blank area and select Display settings.
In the Scale and layout section, adjust the resolution.
Use the Remote Desktop Connection (RDC) tool on your computer to adjust the desktop resolution of the instance.
You cannot adjust the desktop resolution of a Windows instance after you connect to the instance from your computer by using RDC. You must adjust the display settings in RDC on your computer, and connect to the instance by using RDC to apply the new display settings to the instance.
Why do two cursors appear after I log on to a Windows instance by using a VNC management terminal?
If two cursors appear when you use a VNC management terminal to log on to a Windows instance of the g7, c7, or r7 instance family, perform the following steps to modify mouse settings:
Modify mouse settings in Control Panel.
Open Control Panel, set View by to Small icons. Then, click Mouse.
In the Mouse Properties dialog box, click the Pointer Options tab, clear Enhance pointer precision, and then click OK.
Modify mouse settings in the registry.
NoteIf two cursors do not appear in the user logon window, skip this step.
Open Registry Editor.
In the left-side navigation pane, click HKEY_USERS and choose
in the top navigation bar.In the Find dialog box, enter MouseSpeed and click Find Next.
Double-click MouseSpeed. In the Edit String dialog box, set Value data to 0 and click OK.
Use a VNC management terminal to log on to the Windows instance again.
What do I do if a DPDK application cannot be deployed on an ECS instance?
We recommend that you deploy DPDK applications on g5ne instances. For information about the g5ne instance family, see Overview of instance families.
If you deploy a DPDK application on an instance of a sixth-generation (such as g6, c6, or r6) instance family or later, an exception may occur when the DPDK application runs. For example, when you use Pktgen-DPDK to test the packet forwarding rate of the instance, the igb_uio port to which the network interface controllers (NICs) are bound may not be detected and the following error is reported:
EAL: eal_parse_sysfs_value(): cannot open sysfs value /sys/bus/pci/devices/0000:00:06.0/uio/uio0/portio/port0/start
We recommend that you use one of the following methods to resolve the issue:
Upgrade DPDK to DPDK Release 21.05 or later. This method is applicable to x86-based instances. For more information about DPDK versions, visit the official DPDK website.
Replace Userspace I/O (UIO) drivers with Virtual Function I/O (VFIO) drivers. This method is applicable to both x86-based instances and Arm-based instances. For more information, see Replace UIO drivers with VFIO drivers.
Can I upgrade the instance types and other configurations of subscription instances?
Yes, you can upgrade the instance types and other configurations of subscription instances. For more information, see Upgrade the instance types of subscription instances.
Can I upgrade the instance types and other configurations of pay-as-you-go instances?
Yes, but you must stop pay-as-you-go instances before you can upgrade their instance types and other configurations. For more information, see Change the instance type of a pay-as-you-go instance or ModifyInstanceSpec.
How long does it take to upgrade the configurations of an instance?
Subscription instances do not need to be stopped for their instance types to be upgraded. It takes about 15 minutes to upgrade the instance type of a subscription instance.
Pay-as-you-go instances must be stopped for their instance types to be upgraded. It also takes about 15 minutes to upgrade the instance type of a pay-as-you-go instance.
You can upgrade the bandwidth of instances without stopping the instances. The upgrade process takes about 5 minutes.
How is the fee for an instance configuration upgrade calculated?
The fee and the fee calculation method are displayed in the ECS console when you upgrade the instance type or other configurations of an instance. You can also go to the Account Overview page in the Expenses and Costs console to view the fee.
Are my cloud service configurations affected if I upgrade the configurations of ECS instances?
Pay-as-you-go instances must be stopped before their configurations can be upgraded. After you upgrade the configurations of a subscription instance, you must restart the instance for the new configurations to take effect. The upgrade operation interrupts the services that run on the instance for a short period of time. We recommend that you upgrade the configurations of instances during off-peak hours. Instances can seamlessly resume services after their configurations are upgraded without the need to reconfigure their environments.
How do I upgrade ECS resources?
For information about how to upgrade ECS resources, see Overview of instance configuration changes.
ECS instances, except those that use local storage, can have their CPU and memory resources scaled and their bandwidth upgraded while the instances are running. After the post-upgrade configurations of instances take effect, you can also downgrade the configurations of the instances.
Up to 16 data disks can be attached to each ECS instance. You can extend cloud disks. Cloud disks cannot be shrunk after they are extended.
The bandwidth of each ECS instance is measured in Mbit/s and can range from 0 Mbit/s to 200 Mbit/s. You can modify the bandwidth or change the billing method for network usage.
I have upgraded the configurations of an instance but no changes have taken effect. Why?
After you upgrade the configurations of an instance, you must restart the instance by using the ECS console or by calling an API operation for the new configurations to take effect.
After I place an order to upgrade the configurations of an instance, can I cancel the order to restore the instance to its original configurations?
No, after an order to upgrade the configurations of an instance takes effect, the order cannot be canceled and the configurations of the instance are upgraded. If you want to restore the instance to its original configurations, you can downgrade the instance configurations. You are charged for the configuration downgrade.
What do I do if I cannot access a website that runs on an ECS instance?
This issue may occur due to various causes. Resolve the issue based on the causes. For more information, see What do I do if I cannot access a website hosted on an instance?
What do I do if an ECS instance remains in the Starting state for an extended period of time and the Aliyun Assist Service process is disabled or deleted?
Problem description: After an ECS instance was started, it remained in the Starting state for an extended period of time and then automatically stopped. You log on to the operating system of the instance to view system services and find that the Aliyun Assist Service process is deleted or disabled.
Solutions:
If the Aliyun Assist Service process is disabled, perform the following operations:
Change the state of Aliyun Assist Service to Automatic.
Restart the ECS instance.
If the Aliyun Assist Service process is deleted, perform the following operations:
Run the following command to add the Aliyun Assist Service process to the instance:
sc create AliyunService type= "own" start= "auto" binPath= "C:\ProgramData\aliyun\assist\{Version}\aliyun_assist_service.exe -d" tag= "no" DisplayName= "AliyunService"
NoteMake sure you leave a space after the equal signs (
=
).Open the HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\AliyunService key and change
C:\ProgramData\aliyun\assist\{Version}\aliyun_assist_service.exe -d
to"C:\ProgramData\aliyun\assist\{Version}\aliyun_assist_service.exe" -d
.Restart the ECS instance.
How do I use FTP tools in macOS to upload files?
Method 1: Use the Terminal of macOS to upload files
Open the Terminal in macOS or iTerm2 for macOS. To download iTerm2, visit the iTerm2 website. Make sure that you select the correct destination path.
Connect to the FTP server.
Access the destination directory. In Windows, use the working directory as the destination directory. In Linux, switch to the htdocs directory.
Run the put command to upload files.
Method 2: Use a third-party tool to upload files
Download Yummy FTP.
Install Yummy FTP.
Enter the server IP address, username, and password. Set Protocol to Standard (FTP). Set Port to 21 or a different port number that you are using, and do not select an SSH key.
Click Connect.
In the right-side pane, select the destination directory. In Windows, use the current working directory. In Linux, select the htdocs directory. In the left-side pane, select files. Then, click the Upload icon to upload the files.
If you attempt to install Yummy FTP and are prompted with a message similar to "Your security preferences only allow the installation of applications from the Mac App Store and authorized developers", perform the following steps:
Choose
.Click the security lock in the lower-left corner of the window and enter the password of the administrator.
Set Allow apps download from to Anywhere.
Then, you can use Yummy FTP to upload files.
If you have further questions, submit a ticket.
How do I apply for an ICP filing for my domain name after I purchase an ECS instance?
Before you apply for an ICP filing, make sure that your ECS instance meets the ICP filing requirements. For more information, see Check the instance for ICP filing and access information.
An ECS instance cannot load the kernel to start. What do I do?
Problem description: The system did not respond when you selected any option from the GRand Unified Bootloader (GRUB) menu on system startup. After you mounted the LiveCD image to the ECS instance and logged on to the instance, you found that the file system permissions were correct and that message logs did not display exceptions.
Cause: The system was infected by ransomware.
Solution: Back up your data and re-initialize the system.
How do I change the logon password from within an instance?
For information about how to change the logon password from within an instance, see Reset the logon password of an instance from within the instance.
Why am I unable to add sound or video cards to ECS instances?
Alibaba Cloud ECS instances are not multimedia servers and do not provide sound or video card components. Sound or video cards cannot be added to ECS instances.
Can I transfer the unused time of an ECS instance to another ECS instance?
No, the unused time of an ECS instance cannot be transferred to other ECS instances. If you want both flexibility and cost-effectiveness, we recommend that you use a combination of reserved instances and pay-as-you-go instances. For more information, see Overview of reserved instances.
Do ECS instances provide databases by default?
No, ECS instances do not provide databases by default. To use database services, perform one of the following operations:
Deploy your own database.
Purchase ApsaraDB RDS instances in the ApsaraDB RDS console.
Use an Alibaba Cloud Marketplace image to deploy databases.
Can I build a database on an ECS instance?
Yes, you can install database software and configure a database environment on an ECS instance. You can also separately purchase ApsaraDB RDS instances in the ApsaraDB RDS console.
Do ECS instances support Oracle databases?
Yes, ECS instances support Oracle databases. Before you install an Oracle database, we recommend that you perform a performance stress test on the ECS instance to ensure that the instance can satisfy the read and write requirements of the database.
Are public and private IP addresses independent? Can I specify or add IP addresses?
In the classic network, public and private IP addresses are independent of each other. Private IP addresses in the classic network are used for communication between ECS instances and between ECS instances and OSS buckets or ApsaraDB RDS instances. If the public bandwidth of an ECS instance is 0 Mbit/s, no public IP address is assigned to the instance. In normal cases, public and private IP addresses in the classic network do not change. You cannot specify, select, or add IP addresses in the classic network.
In VPCs, NAT gateways map public IP addresses to private IP addresses. You can add IP addresses by specifying or automatically assigning secondary private IP addresses to elastic network interfaces (ENIs).
Can load balancing be implemented for a single ECS instance?
Both Linux and Windows ECS instances can be load-balanced. Make sure that the configurations of ECS instances used as web servers meet the requirements for website code to run. Load balancing can be implemented for as few as one ECS instance. However, we recommend you implement load balancing for two or more ECS instances in an account.
Can I change the region of an ECS instance that I purchased?
No, the regions of purchased ECS instances cannot be changed. To change the region of an ECS instance, you can use the ACS-ECS-CloneInstancesAcrossRegion
public template of CloudOps Orchestration Service to copy the instance to another region. The new instance has the same disk data as the original instance but a different IP address.
Can I adjust the partition size of a purchased disk?
For system security and stability purposes, system disks cannot be repartitioned on Windows or Linux instances. If you use a third-party tool to repartition system disks, unknown exceptions such as system failures and data loss may occur.
If you repartition a data disk, data may be lost. We recommend that you do not repartition data disks.
How do I replace the static public IP address (also called auto-assigned public IP address or system-assigned public IP address) of an ECS instance with an EIP?
To replace the static public IP address of an ECS instance with an EIP, make sure that the instance uses the pay-by-bandwidth billing method for network usage and that you have purchased an EIP. Then, use one of the following methods to replace the static public IP address with the EIP:
Method 1
Change the billing method for network usage from pay-by-bandwidth to pay-by-traffic. For more information, see Public bandwidth.
Convert the static public IP address of the instance into an EIP. For more information, see Convert the static public IP address of an ECS instance in a VPC to an EIP.
Disassociate the EIP that you obtained in the previous step from the instance. For more information, see Disassociate an EIP from a cloud resource.
Associate the new EIP that you purchased with the ECS instance. For more information, see Associate an EIP with an ECS instance.
Method 2
In the OOS console, use the
ACS-ECS-ConvertsPublicIPToNewEIPByInstanceId
template to replace the static public IP address of the ECS instance with an EIP. To perform this operation in the China (Hangzhou) region, access the ACS-ECS-ConvertsPublicIPToNewEIPByInstanceId template in the OOS console and select China (Hangzhou) from the drop-down region list in the top navigation bar.
How do I view subscription ECS instances in all regions in my account?
You can go to the Renewal page to view subscription ECS instances in all regions in your account.
Log on to the ECS console.
In the top navigation bar, choose
.
When can I force stop an ECS instance? What are the consequences?
If an instance cannot be stopped by a graceful shutdown procedure, you can force stop the instance. A forced stop is equivalent to disconnecting the power from the instance and can result in loss of unsaved data.
Why am I unable to reactivate an ECS instance?
You may be unable to reactivate an ECS instance due to one of the following reasons:
You have overdue payments in your account. Pay the outstanding bills and try again.
The system is busy. Try again later.
Resources of the instance type are sold out.
NoteYou can go to the ECS Instance Types Available for Each Region page to view the instance types available in each region.
An ECS instance for which release protection was enabled was automatically released from a scaling group. Why?
After an ECS instance is automatically created in a scaling group, Auto Scaling can release the instance even if you enable release protection for the instance in the ECS console or by calling the ModifyInstanceAttribute operation.
To prevent the ECS instances for which release protection is enabled from being released in scaling groups, you can put the ECS instances into the Protected state in the Auto Scaling console. For more information, see Manually change the status of instances.
How do I test the packet forwarding rate of an instance?
For information about how to test the packet forwarding rate of an instance in most cases, see Test the network performance of an instance. For information about how to test the packet forwarding rate of a g7, c7, or r7 instance, see Use Pktgen to test the network performance of ECS instances.
How do I migrate data between ECS instances?
Perform different operations to migrate data from one instance to another based on whether both instances are deployed in the same region and belong to the same account.
To migrate instance data in the same region and the same account, use the image or snapshot service.
Use the image service to migrate instance data.
Create a custom image from the source instance.
Use the custom image to create an instance or replace the image of an existing instance. For more information, see Create an ECS instance by using a custom image and Replace the operating system (system disk) of an instance.
Use the snapshot service to migrate instance data.
Create snapshots for the disks that are attached to the source instance. For more information, see Create a snapshot.
Create disks from the snapshots. For more information, see Create a disk from a snapshot.
Attach the new disks to the destination instance. For more information, see Attach a data disk.
To migrate instance data across regions in the same account, perform the following steps:
Create a custom image from the source instance.
Copy the custom image to the destination region. For more information, see Copy a custom image.
Use the image copy in the destination region to create an instance or replace the image of an existing instance. For more information, see Create an ECS instance by using a custom image and Replace the operating system (system disk) of an instance.
To migrate instance data across accounts in the same region, perform the following steps:
Create a custom image from the source instance.
Share the custom image to the destination account. For more information, see Share a custom image.
Use the shared image in the destination account to create an instance or replace the image of an existing instance. For more information, see Create an ECS instance by using a custom image and Replace the operating system (system disk) of an instance.
To migrate instance data across regions and accounts, perform the following steps:
Create a custom image from the source instance.
Copy the custom image to the destination region. For more information, see Copy a custom image.
Share the new image (image copy) to the destination account. For more information, see Share a custom image.
Use the shared image in the destination account to create an instance or replace the image of an existing instance. For more information, see Create an ECS instance by using a custom image and Replace the operating system (system disk) of an instance.
For more information, see Use a custom image to replicate ECS instances across regions.
In addition, Server Migration Center (SMC) supports the full migration, incremental migration, batch migration, and VPC-based migration features and allows you to migrate data between ECS instances in the SMC console. For more information, see Migrate data between ECS instances that belong to the same Alibaba Cloud account or different Alibaba Cloud accounts.
What is the AliVulfix process in an ECS instance?
The AliVulfix process is a program that is provided by Alibaba Cloud Security Center. The program scans ECS instances for vulnerabilities.
How do I protect ECS instances against attacks?
ECS instances use Alibaba Cloud Security Center to defend against DDoS attacks. CloudMonitor that is deployed on ECS instances can automatically detect network attacks and scrub suspicious traffic. Alibaba Cloud performs blackhole filtering to protect ECS instances against high-volume attacks. To strengthen the security of ECS instances, we recommend that you install security software and disable ports that are infrequently used.
What security services does Alibaba Cloud provide?
Alibaba Cloud Security Center is backed by the robust data analysis capabilities of the cloud computing platform of Alibaba Cloud and provides a comprehensive set of security services, such as security vulnerability detection, website trojan detection, host intrusion detection, and anti-DDoS protection.
For information about more security services, visit the Alibaba Cloud Security Products page.
How do I check for mining programs on a server?
You can use one of the following methods to check for mining programs on your server. For more information, see the Characteristics of mining programs section in the "Best practices for handling mining programs" topic.
Check whether the CPU of your server is operating as expected.
NoteIf the CPU utilization of your server significantly increases (for example, to 80% or higher) and an unknown process continuously transmits packets to external networks, a mining program is running on your server.
Go to the Alerts page in the Security Center console and check for unhandled alerts of mining programs. If mining programs run on your ECS instance, alerts are generated and need to be handled.
For more information, see Defend against mining programs and Best practices for handling mining programs.
How do I handle mining programs or apply to unlock affected servers?
Handle mining programs
You can handle mining programs or apply to unlock affected servers in the Security Center console or handle mining worms in the Cloud Firewall console. For more information, see Defend against mining programs and Best practices for handling mining programs.
Unlock servers
If a server is locked due to infection by a mining virus or attack, you can apply to unlock the server. To apply to unlock servers, go to the Penetration Test Permission page of the Security Control console.
When you apply to unlock servers, take note of the following items:
You can apply to unlock an individual server only once.
A server that has been unlocked is automatically checked within three days. If a mining program is detected on the server again, the server is locked and cannot be unlocked.
After the server is unlocked, back up the data on the server at your earliest opportunity.
I have renewed an expired Linux instance but I am still unable to access the website hosted on it. What do I do?
Problem description: A Linux ECS instance is in the Stopped state after it expired. After you renewed and started the instance, you still cannot access the website that is hosted on the instance.
Cause: This may be because the website service has not been started.
Solution:
Connect to the instance and run the following command to check whether the website service is started:
#netstat -nltp //Check whether port 80 on the ECS instance is being listened on.
If no information about port 80 is displayed in the command output, the website service has not been started. Run a command to start the website service and relevant services.
In Linux, websites are typically developed based on PHP and MySQL.
In Apache, you need to only start the website service and MySQL.
#/etc/init.d/httpd start //Start the website service. This command is applicable to Apache. #/etc/init.d/mysqld start //Start MySQL.
In NGINX, you must start the website service, PHP, and MySQL.
#/etc/init.d/nginx start //Start the website service. This command is applicable to NGINX. #/etc/init.d/php-fpm start //Start PHP. #/etc/init.d/mysqld start //Start MySQL.
Check again whether the website service is started.
#netstat -nltp //Check whether port 80 on the ECS instance is being listened on.
After the website service is started, access the website again.
If the issue persists, submit a ticket.
How do I activate a Windows ECS instance in a VPC?
To activate a Windows ECS instance in a VPC, you must use a specific Key Management Service (KMS) domain name. For more information, see How do I use KMS domain names to activate a Windows instance in a VPC?.
How do I query, partition, and format the disks of a Linux instance?
You can run the df -h command to check the capacity and usage of disks, and run the fdisk -l command to query disk information. For information about how to partition and format the disks of Linux instances, see Initialize a data disk whose size does not exceed 2 TiB on a Linux instance.
How do I upload files to a Linux instance?
You can use the FTP service to upload files to a Linux instance.
How do I change the owner and owner group of directories and files on a Linux instance?
If the file or directory permissions are not correctly configured on an instance that works as a web server, a 403 error is reported when you access a website hosted on the instance. Before you adjust a file or directory, you must identify the identity under which the file or directory process is running.
You can run the ps and grep commands to query the identities under which processes are running.
You can run the ls -l command to query the owners and owner groups of files and directories.
To change the owners and groups, run the chown command. For example, you can run the chown -R www.www /alidata/www/phpwind/ command to change the owners and groups of all files and directories under the /alidata/www/phpwind directory to the www account.
How do I update software repositories for Linux instances?
You can use an automatic tool to update software repositories for Linux instances. For more information, see Use scripts to automatically update software repositories in Linux instances.
What limits apply to the transfer and change of public IP addresses of ECS instances?
The following limits apply to the transfer and change of public IP addresses of ECS instances:
ECS instances in the classic network
You cannot transfer public IP addresses across accounts.
The public IP address of an instance can be changed within 6 hours after the instance is created, and can be changed up to three times. For more information, see Change the public IP address of an ECS instance.
ECS instances in VPCs
You cannot transfer public IP addresses or EIPs across accounts.
If no public IP address is assigned to an instance, you can associate an EIP with the instance. You can replace the EIP by associating a new EIP with the instance.
If you assign a public IP address to an instance during instance creation, take note of the following limits:
The public IP address of an instance can be changed within 6 hours after the instance is created, and can be changed up to three times. For more information, see Change the public IP address of an instance.
You can convert the public IP address of the instance into an EIP and then replace the EIP. For more information, see Convert the public IP address of an ECS instance in a VPC to an EIP.
If you have further questions, submit a ticket.
Can I access amazon.com from my ECS instance?
You can access amazon.com from your ECS instance if the instance can connect to the Internet.
Why am I unable to access a website hosted outside the Chinese mainland after I log on to my ECS instance?
Websites hosted outside the Chinese mainland can be accessed by your ECS instance only when the websites comply with the laws, regulations, and regulatory requirements of the country or region where your instance is located. Make sure that your ECS instance can connect to the Internet and that the website complies with the preceding laws, regulations, and regulatory requirements.
I cannot purchase more pay-as-you-go instances. What do I do?
If you have reached the maximum number of pay-as-you-go instances that you can purchase, you cannot purchase additional pay-as-you-go instances. For more information, see the "Instance limits" section in Limits. You can log on to the ECS console and click Privileges on the Overview page to view your resource quotas. For more information, see Manage quotas.
How can I view the resource quota?
For more information about how to view the limits and quotas of resources, see Limits.
After a pay-as-you-go instance is stopped manually or automatically due to an overdue payment, am I still charged for the instance?
Instance stop due to an overdue payment: When your account has overdue payments, your pay-as-you-go instances automatically stop and billing for the instances stops. Instances do not always remain in the Stopped state after they are stopped due to overdue payments. For more information, see Pay-as-you-go.
Manual instance stop: You can use the ECS console or call the StopInstance operation to stop a running pay-as-you-go instance. When the instance is stopped, it enters the Stopped state. How stopped pay-as-you-go instances are billed depends on their network types.
VPC: You can enable economical mode for pay-as-you-go instances that reside in VPCs.
If economical mode is enabled, billing for pay-as-you-go instances begins when the instances are created, and billing for some instance resources stops when the instances enter the Stopped state and resumes when the instances are restarted. When a pay-as-you-go instance is stopped in economical mode, billing stops only for the computing resources (vCPUs and memory) and static public IP address of the instance and continues for other resources, such as the disks and EIP. For more information, see Economical mode.
If economical mode is disabled, billing for pay-as-you-go instances continues when they enter the Stopped state.
Classic network: Pay-as-you-go instances in the classic network are billed regardless of whether they are in the Stopped state.
What do I do if an order cannot be placed to change the billing method of an instance from pay-as-you-go to subscription?
You may be unable to place the order due to one of the following reasons:
The instance is in a state that does not support billing method changes. For example, you have an unpaid order for the instance.
Billing method changes are not allowed due to an upcoming scheduled automatic release.
Billing method changes are not allowed because instance information has been changed.
A previous order to change the billing method of the instance has not been paid.
If one of the preceding errors is reported, adjust the instance accordingly.
How long after an order is paid does it take to change the billing method of an instance from pay-as-you-go to subscription?
The billing method of your instance is changed after the order is paid. It can take up to 4 seconds to change the billing method of up to 20 instances. After the change is complete, you can see that the billing method of your instance is changed to Subscription in the ECS console.
What do I do if the billing method of an instance cannot be changed from pay-as-you-go to subscription?
To resolve this issue, submit a ticket.
When I change the billing method of an instance from pay-as-you-go to subscription, does the billing method for network usage of the instance also change?
No, the billing method for network usage of the instance does not change. Only the billing method of instances and disks can be changed from pay-as-you-go to subscription. For information about how to change the billing method for network usage, see Overview of instance configuration changes.
I have an unpaid order to change the billing method of an instance from pay-as-you-go to subscription. If I upgrade the configurations of the instance, is the order still valid?
An order is created when you change the billing method of your instance from pay-as-you-go to subscription. You must pay for the order to complete the change. If you upgrade the configurations of the instance before the order is paid, the order payment cannot be completed because the instance components are different and the original order no longer matches the instance components. If you still want to change the billing method of your instance, you must cancel the unpaid order and place a new order.
What do I do if the billing method of an instance cannot be changed from subscription to pay-as-you-go?
You may be unable to change the billing method of an instance from subscription to pay-as-you-go due to one of the following reasons:
The instance is in a state that does not support billing method changes. For example, you have an unpaid order for the instance.
The instance is in the Expired state.
Billing method changes are not allowed because instance information has been changed. For example, the bandwidth of the instance has been temporarily upgraded.
If one of the preceding errors is reported, adjust the instance accordingly. If the issue persists, submit a ticket.
When I attempt to change the billing method of a disk in an instance, an error message is returned indicating that I have changed the billing method three times. What does this mean?
Each subscription ECS instance can have its configurations downgraded up to three times. Downgrade operations include instance type downgrades, bandwidth downgrades, and the change of the disk billing method from subscription to pay-as-you-go.
Why am I unable to change a pay-as-you-go instance into a subscription instance?
The pay-as-you-go instance that you want to change must meet the following requirements:
The instance type of the instance is not retired. For more information, see Retired instance types.
The instance is not a preemptible instance.
You do not have unpaid orders for the instance.
If you have unpaid orders for the instance, you must pay for the orders or cancel the orders before you can change the billing method of the instance.
The automatic release time is not set for the instance.
If the automatic release time is set for the instance, you must cancel the automatic release of the instance before you change its billing method. For more information, see Release an instance.
The instance is in the Running or Stopped state.
If you placed an order to change the billing method of an instance in the Running or Stopped state and then the instance entered a different state before the payment is completed, the payment fails and the billing method does not change. You can go to the Billing Management console and pay for the order when the instance is in the Running or Stopped state again.
How do I view the expiration time of a subscription instance?
You can log on to the ECS console and go to the Instance page to view the expiration time of your subscription instance in the Billing Method column.
If the Billing Method column does not appear, click the icon in the upper-right corner. In the Instance List Settings dialog box, select Billing Method and then click Continue.