broker: reduce tbon.tcp_user_timeout #4632
Merged
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
Problem: the default time to wait for a TCP ACK from a peer broker before disconnecting is too long. The system default is around 20m, which is a very long time for a free range batch job to be hung up after losing a non-critical node. Use 20s as the default. This can still be overridden via TOML config and broker command line. Update sharness test that checked for a default of 0 (system).
Problem: the default tbon.tcp_user_timeout has changed, but the man page still says it uses the system default. Update man page.
|
I had forgotten (and was reminded by CI) that this attribute is not supported on el7, and my change didn't account for that. Repushed with minor tweaks so that the logic for ensuring it's an error to tune this when it's not tunable still works. |
Codecov Report
@@ Coverage Diff @@
## master #4632 +/- ##
==========================================
+ Coverage 83.39% 83.41% +0.01%
==========================================
Files 411 411
Lines 68792 68796 +4
==========================================
+ Hits 57371 57388 +17
+ Misses 11421 11408 -13
|
|
Restarting bionic builder that failed here: hopefully just a temporary github issue |
|
Thanks, I'll set MWP. |
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Outdated suggestions cannot be applied.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.
Problem: a batch job takes around 20m to notice that a peer has been turned off.
The
tbon.tcp_user_timeoutcan be used to tune this value in the system instance, and we encourage that. But batch jobs get the system default.Since the system default is probably not suitable for flux ever, set our own default. Let's try 20s.