Skip to content

Fixes for rescaling streams with sources #5428

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.

Sign up for GitHub

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

Merged
derekcollison merged 6 commits into from
May 16, 2024
+520 −8
Merged

Fixes for rescaling streams with sources #5428

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
17813e3
Add test to simulate busy stream sources and rescaling
wallyqs May 14, 2024
e52a88d
Add test with busy streams not using sources
wallyqs May 15, 2024
256e3d7
Use atomic to avoid data race
derekcollison May 15, 2024
de073ff
When scaling up and down we keep same leader so no reason to tear eve…
derekcollison May 15, 2024
7491702
Do rolling restart after scale jobs
derekcollison May 16, 2024
1e3dc04
Add skip to test
wallyqs May 16, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 7 additions & 4 deletions server/jetstream_cluster.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3578,7 +3578,6 @@ func (js *jetStream) processClusterUpdateStream(acc *Account, osa, sa *streamAss
var needsSetLeader bool
if !alreadyRunning && numReplicas > 1 {
if needsNode {
mset.setLeader(false)
js.createRaftGroup(acc.GetName(), rg, storage, pprofLabels{
"type": "stream",
"account": mset.accName(),
Expand All @@ -3598,10 +3597,14 @@ func (js *jetStream) processClusterUpdateStream(acc *Account, osa, sa *streamAss
} else if numReplicas == 1 && alreadyRunning {
// We downgraded to R1. Make sure we cleanup the raft node and the stream monitor.
mset.removeNode()
// Make sure we are leader now that we are R1.
needsSetLeader = true
// In case we need to shutdown the cluster specific subs, etc.
mset.setLeader(false)
mset.mu.Lock()
// Stop responding to sync requests.
mset.stopClusterSubs()
// Clear catchup state
mset.clearAllCatchupPeers()
mset.mu.Unlock()
// Remove from meta layer.
js.mu.Lock()
rg.node = nil
js.mu.Unlock()
Expand Down
Loading