New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Following restart node continues to log "Commitlog shutdown complete" messages #11508
Comments
2e1ec36 added shutdown logging to the commitlog. It was probably overlooked that hinted handoff shuts down a commitlog every 10s. At: Line 489 in 7effd4c
I have no idea about hints; I just followed the code up from the log message. So I don't have any fix ideas. cc @xemul |
@michoecho , I probably mis-understood the commitlog call sequence and put the logging into wrong place. The intention was to print something into logs when database shuts down its commitlog on scylla stop/abort. Feel free to remove this log line if you think it's not applicable there. |
I see it in my tests as well. |
@xemul I think you put it in the place you intended. The problem is, the main database commitlog isn't the only commitlog in the program. Hinted handoff reuses the implemention for it's own little transient commitlogs, which are destroyed and recreated (or just shut down and reopened, I don't know how exactly it works) every 10s, and this causes the log spam. |
The intention was for these logs to be printed during the database shutdown sequence, but it was overlooked that it's not the only place where commitlog::shutdown is called. Commitlogs are started and shut down periodically by hinted handoff. When that happens, these messages spam the log. Fix that by adding INFO commitlog shutdown logs to database::stop, and change the level of the commitlog::shutdown log call to DEBUG. Fixes scylladb#11508
A fix (#11536) is already queued. |
The intention was for these logs to be printed during the database shutdown sequence, but it was overlooked that it's not the only place where commitlog::shutdown is called. Commitlogs are started and shut down periodically by hinted handoff. When that happens, these messages spam the log. Fix that by adding INFO commitlog shutdown logs to database::stop, and change the level of the commitlog::shutdown log call to DEBUG. Fixes #11508 Closes #11536 (cherry picked from commit 9b6fc55)
Backported to 5.1. Earlier releases are unaffected. |
Installation details
Kernel Version: 5.15.0-1019-aws
Scylla version (or git commit hash):
2022.2.0~rc1-20220902.a9bc6d191031
with build-id074a0cb9e6a5ab36ba5e7f81385e68079ab6eeda
Cluster size: 5 nodes (i3.4xlarge)
Scylla Nodes used in this run:
OS / Image:
ami-03d21402486bbce67
(aws: eu-west-1)Test:
longevity-mv-si-4days-test
Test id:
371f29c2-5d87-4a5a-a487-a974ebf7edd1
Test name:
enterprise-2022.2/longevity/longevity-mv-si-4days-test
Test config file(s):
Issue description
>>>>>>>
During
RollingConfigChangeInternodeCompression
we change the internode compression settings (duh!) and restart the nodes in the cluster in a rolling fashion.During the shutdown process we get the expected log line informing about the commitlog shutdown:
node 1 db log:
However this message gets logged after the restart every few seconds:
node 1 db log:
This continues on all the db nodes, through 7+ hours of the test, taking up ~85% of the db nodes log output:
<<<<<<<
$ hydra investigate show-monitor 371f29c2-5d87-4a5a-a487-a974ebf7edd1
$ hydra investigate show-logs 371f29c2-5d87-4a5a-a487-a974ebf7edd1
Logs:
Jenkins job URL
The text was updated successfully, but these errors were encountered: