storage: require enterprise license for WAL failover #121165
Conversation
If WAL failover is configured but the user has not provided an enterprise license, WAL failover will refuse to failover and log a warning message every 10 minutes. Epic: none Release note: none
![blathers[crl]](https://avatars.githubusercontent.com/in/59700?s=40&u=373d9101a12edc3847d2fb6ae78e3adcfb4ca41e&v=4){kind=avatar}
|
It looks like your PR touches production code but doesn't add or edit any test code. Did you consider adding tests to your PR? 🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf. |
|
This change is |
|
I'm working on some |
There was a problem hiding this comment.
Reviewable status:
complete! 1 of 0 LGTMs obtained (waiting on @jbowens and @sumeerbhola)
pkg/storage/open.go line 417 at r1 (raw file):
licenseOK := base.CCLDistributionAndEnterpriseEnabled(settings) if !licenseOK && cclWALFailoverLogEvery.ShouldLog() { log.Warningf(context.Background(), "Ignoring WAL failover configuration because it requires an enterprise license.")
Maybe we should log in the version not ok case too? In case someone doesn't finalize the upgrade accidentally
There was a problem hiding this comment.
Reviewed all commit messages.
Reviewable status:
pkg/storage/open.go line 417 at r1 (raw file):
Previously, RaduBerinde wrote…
Maybe we should log in the version not ok case too? In case someone doesn't finalize the upgrade accidentally
do users sometimes intentionally go for days without finalizing?
There was a problem hiding this comment.
Reviewable status:
pkg/storage/open.go line 417 at r1 (raw file):
Previously, sumeerbhola wrote…
do users sometimes intentionally go for days without finalizing?
Probably, if it takes a long time to roll over the nodes.. but then they shouldn't be configuring WAL failover. I definitely heard of cases where they stay in that state unintentionally though.
There was a problem hiding this comment.
TFTRs!
Reviewable status:
pkg/storage/open.go line 417 at r1 (raw file):
Previously, RaduBerinde wrote…
Probably, if it takes a long time to roll over the nodes.. but then they shouldn't be configuring WAL failover. I definitely heard of cases where they stay in that state unintentionally though.
I know of one customer that intends to set the env var fleet-wide, and they won't upgrade all at once. In their case, I think it's expected for WAL failover to be set but the version not yet finalized.
Maybe we should have a more general warning (independent of wal failover) logged at a longer period if the version hasn't been finalized.
|
bors r+ |
If WAL failover is configured but the user has not provided an enterprise license, WAL failover will refuse to failover and log a warning message every 10 minutes.
Epic: none
Release note: none