1. We were holding open FDs longer than we should for consumers causing issues with open FD limits. We now do not hold them open and cap updates a bit better.
2. When doing a stream delete, consumer delete was repeating alot of work that was not necessary, causing longer delays. This has been optimized a bit, still more improvements to be made.
3. We cover all JS under a single export, but that was also trapping GetNext for pull based consumers, and since this was a no-op (is handled at user account level) we were creating alot of garbage service import responses and reverse map entries that had to be garbage collected. We have a fix in to avoind this but still looking for a better one.
4. Still had some lingering references to all exports vs single JS export.
Signed-off-by: Derek Collison <derek@nats.io>
Allow wider scoped filtered subjects.
We introduce a per subject information tracking to filestore to optimize for large mux'd streams and more efficient filtered consumers.
Signed-off-by: Derek Collison <derek@nats.io>
When we had a duplicate detected in R>1 mode we set the skip sequence indicator but were not using that when dealing with underlying store.
Signed-off-by: Derek Collison <derek@nats.io>
We need the consumer state on the stream leader.
Also if we can't find the store yet for a consumer fallback to calculate needsAck.
Signed-off-by: Derek Collison <derek@nats.io>
This allows a domain to be set in the JetStream server block that sets a domain name.
Once set this signals that any leafnode connections should operate as separate JetStream domains.
Each domain <NAME> is accessible via "$JS.<NAME>.API.>", even when connected to the same domain.
Also for mixed mode you can set a jetstream block now that defines a domain but specifies "enabled: false".
Signed-off-by: Derek Collison <derek@nats.io>
If a subject in the system accounts leafnode deny_imports matches $NRG.>
then jetstream is explicitly disconnected and the server can become
leader.
Signed-off-by: Matthias Hanel <mh@synadia.com>
1. When in mixed mode and only running the global account we now will check the account for JS.
2. Added code to decrease the cluster set size if we guessed wrong in mixed mode setup.
Signed-off-by: Derek Collison <derek@nats.io>
We now make sure to sync state of the replicas when a new leader takes over. We also update ack floors regardless of detection on pending list.
Signed-off-by: Derek Collison <derek@nats.io>
If an update was asssigned but failed at the stream group server we would send back the result which would always delete the stream.
Signed-off-by: Derek Collison <derek@nats.io>
We were not properly enforcing server limits. This commit will allow a server to enforce limits but still remain functional even at the JetStream level.
Also fixed a bug for RAFT replay that could cause instability.
Signed-off-by: Derek Collison <derek@nats.io>
The issue was when a state was removed from a server and restarted it would catch up properly.
However upon cluster restart the system could exhibit strange behaviors. This was due to on
catchup not properly creating a meta snapshot when one was received, leaving no meta state to recover.
Signed-off-by: Derek Collison <derek@nats.io>
