Fix problematic ERS cases #8319
Closed
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
Signed-off-by: Jacques Grove <[email protected]>
just a GTID UUID/SID without an offset/interval. Signed-off-by: Jacques Grove <[email protected]>
automatically in some cases (e.g. ERS); however, we were not resetting this sentinel flag in at least two cases: 1) When we have re-parented successfully (setMasterLocked) and 2) When we have successfully promoted a replica (PromoteReplica) Signed-off-by: Jacques Grove <[email protected]>
aquarapid
requested review from
deepthi,
harshit-gangal,
shlomi-noach and
systay
as code owners
will make it easier to debug from log output. Signed-off-by: Jacques Grove <[email protected]>
Signed-off-by: Jacques Grove <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
…future we may extend to _any uk_ Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
| @@ -619,13 +632,16 @@ func (tm *TabletManager) setMasterLocked(ctx context.Context, parentAlias *topod | |||
| return err | |||
| } | |||
| } | |||
| // Clear replication sentinel flag for this replica | |||
| tm.replManager.setReplicationStopped(false) | |||
There was a problem hiding this comment.
Can you please explain why setting to false? It seems to me like if anything we should set to true, because we're replicating up to a certain point, then stopping, correct?
There was a problem hiding this comment.
The logic here is that "false" means that "replication is no longer stopped on purpose; please restart if necessary"; while "true" means "replication is stopped on purpose, please do not restart".
There was a problem hiding this comment.
Got it! Thanks for clarifying, the function name is confusing.
| @@ -752,6 +770,10 @@ func (tm *TabletManager) PromoteReplica(ctx context.Context) (string, error) { | |||
| return "", err | |||
| } | |||
|
|
|||
| // Clear replication sentinel flag for this master, | |||
| // or we might block replication the next time we demote it | |||
| tm.replManager.setReplicationStopped(false) | |||
There was a problem hiding this comment.
Why is a MASTER tablet marked with false? I'd again think we need true here?
There was a problem hiding this comment.
Same logic as above; we want to remove the on-disk flag that will prevent vttablet from "fixing" replication at some later point in time, if this master ever becomes a replica again.
| delete(differenceSet, sid) | ||
| } else { | ||
| differenceSet[sid] = diffIntervals | ||
| } |
Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Shlomi Noach <[email protected]>
Signed-off-by: Malcolm Akinje <[email protected]>
Signed-off-by: Malcolm Akinje <[email protected]>
Signed-off-by: Malcolm Akinje <[email protected]>
Signed-off-by: Malcolm Akinje <[email protected]>
Signed-off-by: Malcolm Akinje <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
CI: Skip TestConsolidatorMemoryLimits, refactor e2e test cluster setup to reduce vreplication e2e test flakiness
Signed-off-by: Andres Taylor <[email protected]>
Signed-off-by: Vicent Marti <[email protected]>
Add release notes for all the versions
Since Go has been upgraded to 1.16, setting GO111MODULE is no longer needed as it is now equivalent to the current default. Signed-off-by: Dirkjan Bussink <[email protected]>
…hecksum-rbr Ignore SBR statements from pt-table-checksum
This adds the `-trimpath` option so that the release binaries don't include builder specific paths and are independent from the path where they have been built. Signed-off-by: Dirkjan Bussink <[email protected]>
With recent Go versions, GOPATH has a built in fallback to use `~/go` when the environment variable is not set. This also works for these pre commit hooks. With this change it's possible to commit without setting GOPATH at all and having it fall back to the defaults that Go itself provides. Signed-off-by: Dirkjan Bussink <[email protected]>
Signed-off-by: Florent Poinsard <[email protected]>
Signed-off-by: Harshit Gangal <[email protected]>
Small build improvements
gomod: do not replace GRPC
…r cross-shard queries for Gen4 Signed-off-by: Harshit Gangal <[email protected]>
Signed-off-by: Andres Taylor <[email protected]>
Incrementing to release 12.0.0-SNAPSHOT
Gen4: order by and enhanced scoping information for table exprs
…t_conns [grpctmclient] Add support for (bounded) per-host connection reuse
…tidset Signed-off-by: Jacques Grove <[email protected]>
aquarapid
requested review from
derekperkins,
dkhenry,
doeg,
GuptaManan100 and
rohit-nayak-ps
as code owners
|
Going to close this and do a new (clean) PR, this is now polluted with merge commits. |
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.
do_not_replicateflag on disk in some cases #8333 where we do not clear thedo_not_replicatesentinel flag that we create during ERS (and other steps) to ensure that replication reporter does not automatically restart replication after we have (intentionally) stopped it.