Skip to content
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

Cherry-picks for 2.10.17-RC.4 #5550

Merged
merged 12 commits into from
Jun 17, 2024
Merged

Cherry-picks for 2.10.17-RC.4 #5550

merged 12 commits into from
Jun 17, 2024

Conversation

derekcollison and others added 12 commits June 17, 2024 09:41
…m sequence is consistent. (#5528)

For a leader, it will skip msgs at the end to eof and remember o.sseq
for the next getNextMsg call.
But this will report stream delivered different for the leader vs the
followers.

Also if a consumer leader processed an ack ahead of delivered it would
sync, but followers would not.
This makes the behavior consistent between leaders and followers.

Signed-off-by: Derek Collison <[email protected]>

---------

Signed-off-by: Derek Collison <[email protected]>
Make these consistent from leaders and followers regardless of store type.
Fixed a bug that would have followers stores inherit redelivered consumer delivered sequence which would break ack gap fill.

Signed-off-by: Derek Collison <[email protected]>
Signed-off-by: Derek Collison <[email protected]>
…eaving of state.

With skipMsgs calling ProposeDirect, if the prop queue was not empty this would place the skip entry ahead of the other entries.
This would result in sequence mismatch and stalled processing.

Bug was we should not bypass the internal prop ipq for the nrg. This corrects it and renames to ProposeMulti.

Signed-off-by: Derek Collison <[email protected]>
…rge number of messages.

When we delete a consumer we need to simulate acking all messages from the ackfloor to the last sequence in the stream.
If the consumer is new and only consumed and acked and few messages, or none, this could be costly and cause operational pauses for the parent stream.

If the number of sequences we would need to process is obver a threshold we will let the checkInterestState() do the cleanup in a separate Go routine.

Signed-off-by: Derek Collison <[email protected]>
…5548)

For subjects `$JS`, `$JS.API`, `$JSC` and `$SYS` subjects, If NoAck is
true they are now allowed, otherwise they will not be allowed.

This allows proper setup of audit streams for production use cases.

Signed-off-by: Derek Collison <[email protected]>

---------

Signed-off-by: Derek Collison <[email protected]>
We often find ourselves without good visibility into what's really going
on in the Raft layer. This endpoint dumps quite a substantial amount of
internal Raft node state.

Filters include:

* `?acc=ACCNAME` to filter by account (defaults to system account if not
specified)
* `?group=GROUP` to show only specific groups

Signed-off-by: Neil Twigg <[email protected]>

Signed-off-by: Neil Twigg <[email protected]>
…ove restored stream.

We did not remove it which could cause bad healthz reports on single servers for instance.

Signed-off-by: Derek Collison <[email protected]>
@neilalexander neilalexander requested a review from a team as a code owner June 17, 2024 08:51
Copy link
Member

@wallyqs wallyqs left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@wallyqs wallyqs merged commit e8d6600 into release/v2.10.17 Jun 17, 2024
4 checks passed
@wallyqs wallyqs deleted the neil/21017rc4 branch June 17, 2024 15:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants