HDFS-17368. HA: Standby should exit safemode when resources are available. #6518
Conversation
This file contains 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
|
💔 -1 overall
This message was automatically generated. |
|
💔 -1 overall
This message was automatically generated. |
|
@Hexiaoqiao @tasanuma @aajisaka Hi~ sir. Could you please help me review this PR when you are free? Thanks. |
|
💔 -1 overall
This message was automatically generated. |
|
The failed unit test seems unrelated to the change. |
Hexiaoqiao
reviewed
Mar 7, 2024
| @@ -1582,6 +1582,10 @@ void startStandbyServices(final Configuration conf, boolean isObserver) | |||
| standbyCheckpointer = new StandbyCheckpointer(conf, this); | |||
| standbyCheckpointer.start(); | |||
| } | |||
| if (isNoManualAndResourceLowSafeMode()) { | |||
| LOG.info("Standby should not enter safe mode when resources are low, exiting safe mode."); | |||
| leaveSafeMode(false); | |||
There was a problem hiding this comment.
It is reasonable at first glance, not think carefully, any cases to trigger Standby leave safemode untimely? Thanks.
There was a problem hiding this comment.
I reused the logic from HDFS-17231, and I believe there is no issue. HDFS-17231 enables the ANN to automatically exit ResourceLowSafeMode.
At the same time, I noticed that the 'leaveSafeMode(false)' method also exits 'StartupSafeMode'. I'm not sure if this is an issue; I mentioned this phenomenon in HDFS-17402.
If necessary, I can fix it.
There was a problem hiding this comment.
The mention in "HDFS-2915" is that SNN should not enter resource low safe mode. NNRM thread removed from SNN.
There was a problem hiding this comment.
Consider the following scenario:
Initially, nn-1 is active and nn-2 is standby. The insufficient resources of both nn-1 and nn-2 in dfs.namenode.name.dir, the NameNodeResourceMonitor detects the resource issue and puts nn01 into safemode.
At this point, nn-1 is in safemode (ON) and active, while nn-2 is in safemode (OFF) and standby.
After a period of time, the resources in nn-2's dfs.namenode.name.dir recover, triggering failover.
Now, nn-1 is in safe mode (ON) and standby, while nn-2 is in safe mode (OFF) and active.
Afterward, the resources in nn-1's dfs.namenode.name.dir recover.
However, since nn-1 is standby but in safemode (ON), it unable to exit safe mode automatically.
There was a problem hiding this comment.
Great. Got it. Please check if the failed unit test is related with this changes. Others look good to me.
There was a problem hiding this comment.
Thanks for your review. I passed the failed unit test locally.
|
cc @zhangshuyan0 any more suggestions? |
Hexiaoqiao
approved these changes
Mar 25, 2024
Hexiaoqiao
changed the title
|
Committed to trunk. Thanks @zhuzilong2013 for your contributions! |
|
Thanks @Hexiaoqiao for your review and merge~ |
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.
Description of PR
Refer to HDFS-17368.
The NameNodeResourceMonitor automatically enters safemode when it detects that the resources are not suffcient. NNRM is only in ANN. If both ANN and SNN enter SM due to low resources, and later SNN's disk space is restored, SNN willl become ANN and ANN will become SNN. However, at this point, SNN will not exit the SM, even if the disk is recovered.
Consider the following scenario:
If SNN is detected to be in SM(because low resource), it will exit.
How was this patch tested?
Test in a production environment
For code changes:
LICENSE,LICENSE-binary,NOTICE-binaryfiles?