Deprecate ingester.ring.final-sleep in favor of shutdown-delay #3431
Conversation
Deprecate ingester ring specific option for delaying shutdown after a SIGTERM. Instead use the `shutdown-delay` option which can be added to any component (it is not ingester or ring specific). See #3298 Signed-off-by: Nick Pillitteri <nick.pillitteri@grafana.com>
56quarters
force-pushed
the
56quarters/deprecate-final-sleep
branch
from
9da0b4a
to
2d36459
Compare
| @@ -18,8 +18,6 @@ ingester_client: | |||
|
|
|||
| ingester: | |||
| ring: | |||
| # We want to start immediately. | |||
There was a problem hiding this comment.
am I misunderstanding what final_sleep does or does this comment not make sense?
There was a problem hiding this comment.
I think the comment was just wrong. Maybe it used to apply to join_after?
|
Shutdown delay and "final sleep" serve different purpose. I don't think we should proceed with this PR. Shutdown delay -- delays shutdown. It is implemented as "sleep" after SIGTERM is received, but before shutdown of Mimir starts. In other words shutdown delay delays initiation (!) of shutdown, not termination of process after shutdown actions have been performed. Final sleep is applied after ring's lifecycler has finished with chunk transfer (not used by Mimir) and flushing of data to storage (optionally used by Mimir). This is done as part of shutdown actions (ie. after shutdown delay has expired). It is used to delay stopping of Mimir process so that we can still scrape final metrics from it. |
|
That makes sense to me. Thank you for catching this @pstibrany! |
What this PR does
Deprecate ingester ring specific option for delaying shutdown after a SIGTERM. Instead use the
shutdown-delayoption which can be added to any component (it is not ingester or ring specific).Signed-off-by: Nick Pillitteri nick.pillitteri@grafana.com
Which issue(s) this PR fixes or relates to
See #3298
Checklist
CHANGELOG.mdupdated - the order of entries should be[CHANGE],[FEATURE],[ENHANCEMENT],[BUGFIX]