When an account is switched to interest-only mode due to no interest,
it was not possible to switch that account more than once. But the
function switchAccountToInterestMode() that triggers a switch could
possibly doing it more than once. This should not cause problems
but increased the number of traces in a big super cluster.
Also fixed some flappers and a data race.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
- All writes will now be done by the writeLoop, unless when the
writeLoop has not been started yet (likely in connection init).
- Slow consumers for non CLIENT connections will be reported but
not failed. The idea is that routes, gateway, etc.. connections
should stay connected as much as possible. However if a flush
operation times out and no data at all has been written, the
connection will be closed (regardless of type).
- Slow consumers due to max pending is only for CLIENT connections.
This allows sending of SUBs through routes, etc.. to not have
to be chunked.
- The backpressure to CLIENT connections is increased (up to 1sec)
based on the sub's connection pending bytes level.
- Connection is flushed on close from the writeLoop as to not block
the "fast path".
Some tests have been fixed and adapted since now closeConnection()
is not flushing/closing/removing connection in place.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Updated all tests that use "async" clients.
- start the writeLoop (this is in preparation for changes in the
server that will not do send-in-place for some protocols, such
as PING, etc..)
- Added missing defers in several tests
- fixed an issue in client.go where test was wrong possibly causing
a panic.
- Had to skip a test for now since it would fail without server code
change.
The next step will be ensure that all protocols are sent through
the writeLoop and that the data is properly flushed on close (important
for -ERR for instance).
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Running test suite on a Windows VM, I notice several failures.
Updated the compute of the RTT to be at least 1ns. I think that
this is just an issue with the VM I am running, but that change
will have no impact for normal situations (since setting the rtt
to the very minimum duration (1ns) instead of 0) and will prevent
some tests from failing.
Because of those same timer granularity issues, I had to add some
delays between some actions in order for time.Sub()/Since() to
actually report something more than 0.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
This is needed for mapped gateway replies. We had used an extra
token when implementing the new prefix, but it was then removed,
but the leafnode subscription on _GR_.*.*.*.> was not updated.
We now subscribe on _GR_.>
There was a test that was passing because we were using inboxes
that caused the pattern to match. Replaced with single token
reply so that it would have caught this bug.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
- Risk of deadlock when checking if issuer claim are trusted. There
was a RLock() in one thread, then a request for Lock() in another
that was waiting for RLock() to return, but the first thread was
then doing RLock() which was not acquired because this was blocked
by the Lock() request (see e2160cc571)
- Use proper account/locking mode when checking if stream/service
exports/signer have changed.
- Account registration race (regression from https://github.com/nats-io/nats-server/pull/890)
- Move test from #890 to "no race" test since only then could it detect
the double registration.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
We had too much special processing, so reduced to a single wildcard
which will propagate across routes and gateways and is consistent
with gateway handling of globally routed subjects and timeouts.
Signed-off-by: Derek Collison <derek@nats.io>
Changed code on Windows to not use svc code if running in interactive
mode. The original code was running svc.debug.Run() which uses service
code (Execute()) but from the command line. We don't need that.
Also reduced salt on bcrypt password for a config file that started
to cause failures due to test taking too long to finish.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
- New prefix that includes origin server for the request
- Mapping done if request is service import or requestor has
recent subscription
- Subscription considered recent if less than 250ms
- Destination server strip GW prefix before giving to client
and restore when getting a reply on that subject
- Mapping removed aftert 250ms
- Server rejects client publish on "$GNR." (the new prefix)
- Cluster and server hash are now 8 chars long and from base 62
alphabets
- Mapped replies need to be sent to leafnode servers due to race
(cluster B sends RS+ on GW inbound then RMSG on outbound, the
RS+ may be processed later and cluster A may have given message
to LN before RS+ on reply subject. So LN needs to accept the
mapped reply but will strip to give to client and reassemble
before sending it back)
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Allow disabling of short first ping timer for clients.
Adjust names so that full test suite results are aligned.
Removed the account lookup, we use sync.Map but also a no-lock cache.
Signed-off-by: Derek Collison <derek@nats.io>
Ivan had the idea of using the CONNECT to establish a first estimate of RTT
without additional PING/PONGs.
Signed-off-by: Derek Collison <derek@nats.io>
If a client RTT for a requestor is longer than a service RTT, the requestor latency was often zero.
We now wait for the RTT (if zero) before sending out the metric.
Signed-off-by: Derek Collison <derek@nats.io>
This is achieved by subscribing to a unique subject. If the LS+
protocol is coming back for the same subject on the same account,
then this indicates a loop.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
1. Accounts will show up in connection info if auth=1.
2. You can filter by user (?auth=1&user=ivan) or account (?auth=1&acc=eng)
Signed-off-by: Derek Collison <derek@nats.io>
This is the first pass at introducing exported services to the system account for generally debugging of blackbox systems.
The first service reports number of subscribers for a given subject. The payload of the request is the subject, and optional queue group, and can contain wildcards.
Signed-off-by: Derek Collison <derek@nats.io>
In updateRouteSubscriptionMap(), when a queue sub is added/removed,
the code locks the account and then the route to send the update.
However, when a route is accepted and the subs are sent, the
opposite (locking wise) occurs. The route is locked, then the account.
This lock inversion is possible because a route is registered (added
to the server's map) and then the subs are sent.
Use a special lock to protect the send, but don't hold the acc.mu
lock while getting the route's lock.
The tests that were created for the original missed queue updates
issue, namely TestClusterLeaksSubscriptions() and
TestQueueSubWeightOrderMultipleConnections() pass with this change.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Will now breakout the internal NATS latency to show requestor client RTT, responder client RTT and any internal latency caused by hopping between servers, etc.
Signed-off-by: Derek Collison <derek@nats.io>
With Go 1.12 (strangely was not able to reproduce with Go 1.11)
the test TestRouteNoCrashOnAddingSubToRoute() would frequently
locks up and consume all avail CPUs on the machine. Running
this test with GOMAXPROCS=2 you would see server.test CPU usage
pegged at 200% (assuming you have at least 2 CPUs).
The reason was that the writeLoop was spinning because another
routine was already in flushOutbound() and stack trace would
show that it was stuck in system calls. It seems that even though
the writeLoop does release the lock but grab it right away was
not allowing the syscall to complete.
So decided to put back the unlock/gosched/lock back in flushOutbound()
when flag is already set, but then protect the closeConnection()
with its own flag (similar to clearConnection) to not re-introduce
issue fixed in #1092.
Had to fix the benchmark test RoutedInterestGraph because after a
route is accepted, the initial PING will be sent after 1sec which
was breaking this test.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
When a leaf or route connection is created, set the first ping
timer to fire at 1sec, which will allow to compute the RTT
reasonably soon (since the PingInterval could be user configured
and set much higher).
For Route in PR #1101, I was sending the PING on receiving the
INFO which required changing bunch of tests. Changing that to
also use the first timer interval of 1sec and reverted changes
to route tests.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
- On startup, verify that local account in leafnode (if specified
can be found otherwise fail startup).
- At runtime, print error and continue trying to reconnect.
Will need to decide a better approach.
- When using basic auth (user/password), it was possible for a
solicited Leafnode connection to not use user/password when
trying an URL that was discovered through gossip. The server
now saves the credentials of a configured URL to use with
the discovered ones.
Updated RouteRTT test in case RTT does not seem to be updated
because getting always the same value.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
Added the RTT field to each route reported in routez.
Ensure that when a route is accepted, we send a PING to compute
the first RTT and don't have to wait for the ping timer to fire.
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
