To improve cluster stability, we recommend that you update ack-node-problem-detector to V1.2.8 or later.
Background information
By default, the check_fd inspection item is enabled for ack-node-problem-detector versions earlier than V1.2.8. When the check_fd process reads data from a large kernel, specific bugs of the current kernel version may be triggered. This may cause the number of zombie processes to increase and lead to system failures.
Update ack-node-problem-detector
To avoid the issues that may occur in ack-node-problem-detector versions earlier than V1.2.8, we recommend that you update ack-node-problem-detector to V1.2.8 or later for your cluster if the OS kernel meets the requirements. This disables the check_fd inspection item and prevents the check_fd process from reading data from a large kernel. For more information about the release notes of ack-node-problem-detector, see ack-node-problem-detector.
Fix kernel bugs.
Log on to the ACK console and click Clusters in the left-side navigation pane.
On the Clusters page, click the name of a cluster and choose in the left-side navigation pane.
On the Logs and Monitoring tab of the Add-ons page, find ack-node-problem-detector and click Upgrade.
NoteIf the Upgrade button is not displayed, the current version of ack-node-problem-detector is updated to the latest.
In the Update message, click OK.