The server would reset its INFO's TLSRequired to the presence
of a TLS configuration without checking for the allow_non_tls
option.
Resolves#3581
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
This is based of @neilalexander PR #3558.
It ensures that the timer is reset/canceled on configuration
update (by the leader only).
Fixed also the issue with a super-cluster where the delete timer
would always be reset at every gateway interval check.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Originally, only solicited routes were retried in case of a disconnect,
but that was before gossip protocol was introduced. Since then, two
servers that connect to each other due to gossip should retry to
reconnect if the connection breaks, even if the route is not explicit.
However, server will retry only once or more accurately, ConnectRetries+1.
This PR solves the issue that the reconnect attempt was not initiated
for a "solicited route" that was not explicit.
Maybe related to #3571
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
If we had to flush while loading in subject info we would fail there and not properly rebuild state as we did on a failure for flushPending.
Signed-off-by: Derek Collison <derek@nats.io>
When a server was restarted and expired messages, but the leader had a snapshot that
still had the old messages we would reset complete follower stream state, this fix
just skips over the expired as we prepare the request to the leader.
Resolves#3516
Signed-off-by: Derek Collison <derek@nats.io>
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
If an ephemeral was given a name by the user, if the consumer leader
was then shutdown, the ephemeral would be migrated using a server
generated new name instead of keeping the user given name.
Also, in some cases the migration would not even occur. This was
likely due to the fact that RAFT node(s) were shutdown prior to
the ephemeral migration code was invoked.
Resolves#3550
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
If the server is unable to write to disk, there is a failure mode that
results in the fss map being nil when writing message block data to it,
resulting in a panic. This fix checks whether an error is returned by
the function that loads the fss map, and if so, returns an error to
prevent future assignment.
A simple configuration like this:
```
...
mappings = {
foo: bar
}
mqtt {
port: 1883
}
```
would cause an MQTT subscription on "bar" to not receive messages
published on "foo".
In otherwords, the subject transformation was not done when parsing
a PUBLISH packet.
This PR also handles the case of service imports where transformation
occurs after the initial publish parsing.
Resolves#3547
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
If the user sends a CONSUMER.CREATE request with a configuration that
specifies the name that the user wants for the ephemeral consumer,
this would not work on cluster mode, that is, the server would still
pick a name instead of using the provided one.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
If a stream is created or updated with a negative replicas count,
and error is now returned. Same for consumers.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
While going over message blocks, when some blocks were removed there
was a risk that a block is ignored, leaving some messages around.
Resolves#3528
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Originally, createRaftGroup() would not hold the jetstream's lock
for the whole duration. But some race reports made us change
this function to keep the lock for the whole duration. A test
called TestJetStreamClusterRaceOnRAFTCreate() was demonstrating
the race between "consumer info" request handling and createRaftGroup
code. Since then, the race has been fixed, so this PR restores
the more fine-grained locking inside createRaftGroup.
Resolves#3516
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
There was an observed degradation (around 5%) for large fan out in
v2.9.0 compared to earlier release. This is because we added
accounting of the in/out messages for the account, which result
in 4 atomic operations, 2 for in and 2 for out, however, it means
that for a fan-out of say 100 matching subscriptions, it is now
2 + 2 * 100 = 202.
This PR rework how the stats accounting is done which removes
the regression and even boost a bit the numbers since we are
doing the server stats update as an aggregate too.
There are still degradation for queues and no-sub at all that
need to be looked at.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Updating a consumer configuration from say R3 to R1 would work
but no response was received by the client sending the request.
Resolves#3493
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
