Fix DiskThresholdDecider average disk usage with huge filesystems #100599
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
tlrx
added
>bug
:Distributed Coordination/Allocation
elasticsearchmachine
added
the
Team:Distributed (Obsolete)
|
Pinging @elastic/es-distributed (Team:Distributed) |
|
Hi @tlrx, I've created a changelog YAML for you. |
| @@ -393,6 +399,9 @@ public Decision canRemain(IndexMetadata indexMetadata, ShardRouting shardRouting | |||
| if (indexMetadata.ignoreDiskWatermarks()) { | |||
| return YES_DISK_WATERMARKS_IGNORED; | |||
| } | |||
| if (useAverageDiskUsage(node, usages) == false) { | |||
| return YES_AVERAGE_DISK_USAGE_UNAVAILABLE; | |||
| } | |||
There was a problem hiding this comment.
This is added before every call getDiskUsage that might compute the average.
Is there a value of making getDiskUsage return null if overflow happens and have it as a signal to return above decision? This way the addition of total and free could be performed only once (in getDiskUsage) vs twice (in getDiskUsage and the check above).
| ); | ||
| var decider = new DiskThresholdDecider( | ||
| Settings.EMPTY, | ||
| new ClusterSettings(Settings.EMPTY, ClusterSettings.BUILT_IN_CLUSTER_SETTINGS) |
There was a problem hiding this comment.
Nit:
Suggested change
| new ClusterSettings(Settings.EMPTY, ClusterSettings.BUILT_IN_CLUSTER_SETTINGS) | |
| ClusterSettings.createBuiltInClusterSettings() |
Comment on lines
+141
to
+143
| .add(DiscoveryNodeUtils.builder("node_0").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) | ||
| .add(DiscoveryNodeUtils.builder("node_1").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) | ||
| .add(DiscoveryNodeUtils.builder("node_2").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) |
There was a problem hiding this comment.
NIT:
Suggested change
| .add(DiscoveryNodeUtils.builder("node_0").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) | |
| .add(DiscoveryNodeUtils.builder("node_1").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) | |
| .add(DiscoveryNodeUtils.builder("node_2").roles(new HashSet<>(DiscoveryNodeRole.roles())).build()) | |
| .add(DiscoveryNodeUtils.create("node_0")) | |
| .add(DiscoveryNodeUtils.create("node_1")) | |
| .add(DiscoveryNodeUtils.create("node_2")) |
I believe all roles are added by default so no need to manually set them
elasticsearchmachine
added
v9.1.0
Team:Distributed Coordination
|
Pinging @elastic/es-distributed-obsolete (Team:Distributed (Obsolete)) |
|
Pinging @elastic/es-distributed-coordination (Team:Distributed Coordination) |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
DiskThresholdDecidermay compute an average disk usage if the real disk usage of a node is unknown. This average disk usage computes the average total disk space and free disk space among all nodes, but that can break if nodes are using huge filesystems or if the sum of total/free bytes exceeds Long.MAX_VALUE (we've seen total bytes being negative in some clusters).This change introduces an early check in
DiskThresholdDeciderto detect such situation and treat it as if nodes disk usages were unknown, returning aYESdecision for allocation. This is the less impactful solution I can find for this bug but I'm open to other suggestions.