#Fixing compliance-clock?

1 messages · Page 1 of 1 (latest)

shrewd pendant
#

Not sure how this happened, but we seem to have two different compliance-clocks on two difference HA-Pairs in the same cluster...
`s01::> compliance-clock show
(snaplock compliance-clock show)
Node ComplianceClock Time

s01-01 Wed Nov 20 03:07:08 CET 2024 +01:00
s01-02 Wed Nov 20 02:58:45 CET 2024 +01:00
s01-03 Wed Nov 20 10:07:44 CET 2024 +01:00
s01-04 Wed Nov 20 10:07:44 CET 2024 +01:00
4 entries were displayed.`

Is there a way to "fix" this? The nodes 03-04 have a clock that matches the current NTP time whereas 01-02 is out of time somehow... Sadly systems 01+02 has the most volumes with snaplocked snapshots so it's not easy to "empty" them...

subtle crypt
#

Your first two nodes most likely had a downtime of ~7hours at some point in time, that's why they are behind the current time.

bold shale
#

it should not be an issue. at most it keeps snaplocked files locked for 7 hours longer than intended

shrewd pendant
#

Yes it's just because of my OCD 😉 If I wanted to sync them up, I would need to empty a node right? (or the HA pair?) and does this have any effect when moving volumes between the nodes?

bold shale
#

I don't think emptying a node would be enough, I guess you wuold have to completely remove the node from the cluster, wipe it (including all aggregates), and re-join

#

alternatively, you could shut down nodes 3 and 4 for 7 hours each, then they would at least be in sync again 😂

shrewd pendant
#

Pretty sure it if you try to remove the compliance clock, it tells you that you have locked volumes and you need to "remove" that before you can remove the compliance clock... so maybe it could be done by just relocating the aggregates from one node to the other... 🙂 ...or maybe I should just leave it alone 😉

#

`s01::*> compliance-clock initialize -node s01-01 -force true
(snaplock compliance-clock initialize)

Error: command failed: ComplianceClock re-initialization failed. ComplianceClock re-initialization requires all nodes in the cluster to be healthy, all volumes are in online state, no
volumes are present in the volume recovery queue and there are no SnapLock volumes or volumes with "snapshot-locking-enabled" set to true or S3 buckets with object locking
enabled.`

bold shale
#

no the compliance clock is in the aggregates. you need to move all volumes off, then wipe everything. that should reset the clock on the nodes and when you re-join them it gets the correct time again. Question is: is that worth it just to satisfy your OCD 😄

shrewd pendant
bold shale
#

😄

shrewd pendant
#

but is there an issue when moving volumes (with locked snapshots) between nodes with different compliance clocks?

#

.....I will find a small volumes and test it I guess

bold shale
shrewd pendant