gogrlx/nats-server
1
0
Fork 0
mirror of https://github.com/gogrlx/nats-server.git synced 2026-04-16 19:14:41 -07:00
Code Issues Packages Projects Releases Wiki Activity
4,996 Commits 30 Branches 48 Tags
052bb7ca54280873cf77ab1963ae2f063847c95b
Commit Graph

4996 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Derek Collison
052bb7ca54 Merge to fix conflicts 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-21 08:16:51 -07:00
Derek Collison
a6f95e9886 Merge pull request #2550 from nats-io/issue-2548 
[FIXED] #2548
 2021-09-21 08:06:04 -07:00
Derek Collison
de851e513f Fix for #2548 
Replicated durable consumers that were backed by a memory store were bypassing snapshotting which also did compaction of the raft WAL.
This change adapts for memory store backed consumers by compacting the raft WAL directly on snapshot logic.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-21 08:02:11 -07:00
Derek Collison
63c242843c Avoid panic if WAL was truncated out from underneath of us. 
If we were leader stepdown as well.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-21 07:26:03 -07:00
Derek Collison
12bb46032c Fix RAFT WAL repair. 
When we stored a message in the raft layer in a wrong position (state corrupt), we would panic, leaving the message there.
On restart we would truncate the WAL and try to repair, but we truncated to the wrong index of the bad entry.

This change also includes additional changes to truncateWAL and also reduces the conditional for panic on storeMsg.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-20 18:41:37 -07:00
Ivan Kozlovic
e69c6164a9 Merge pull request #2547 from nats-io/travis_coveralls 
Run coverage, but push to coverall only after_success
 2021-09-20 12:03:17 -06:00
Ivan Kozlovic
d154d646d4 Run coverage, but push to coverall only after_success 
This will ensure that we get the matrix run green before pushing
results to coveralls, but also that if for some reason coveralls
has a problem, this is not making our build status red.

Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-20 12:02:14 -06:00
Matthias Hanel
cf6715bc19 Merge pull request #2539 from nats-io/reserved_memory 
Exposing reserved memory in jsz/varz
 2021-09-20 12:55:07 -04:00
R.I.Pienaar
3be0b23a3e Merge pull request #2546 from ripienaar/negative_dupe_window_protection 
protect against negative dupe window via negative max age
 2021-09-20 18:09:19 +02:00
Derek Collison
83825a2ae5 Merge pull request #2545 from nats-io/fs-stable 
Stabilize filestore, bug fixes, test tweaks.
 2021-09-20 08:59:56 -07:00
R.I.Pienaar
34b5a11e20 protect against negative dupe window via negative max age 
Signed-off-by: R.I.Pienaar <rip@devco.net>
 2021-09-20 17:34:48 +02:00
Derek Collison
c764da328c Make sure to set last sequence on recovery, check for ddloaded on last msg id 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-20 07:11:15 -07:00
Derek Collison
edac5992b8 Bump version 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 14:02:00 -07:00
Derek Collison
cecb6246f2 Test tweaks 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 13:28:37 -07:00
Derek Collison
cfb69bbd1f Only rebuild dedupe state as needed versus always on startup. 
Was penalizing large non-dedupe streams on startup.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 13:27:26 -07:00
Derek Collison
08b498fbda Log error on write errors 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 12:14:31 -07:00
Derek Collison
d79a00499d Remove extra unlock that causes panics 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 12:13:37 -07:00
Derek Collison
4283358dcd Improvments to writeIndexInfo logic and managing open FDs. 
Also hold lock while doing sync and optionally close FDs if idle.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-19 11:45:16 -07:00
Derek Collison
7a4c904761 Improvements to cache management. 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-18 15:21:12 -07:00
Derek Collison
3bfeff67f2 Merge pull request #2540 from nats-io/update-go-mod 
Update nats.go version in go.mod
 2021-09-16 16:52:11 -07:00
Waldemar Quevedo
0bef39ab5b Update nats.go version 
Signed-off-by: Waldemar Quevedo <wally@synadia.com>
 2021-09-16 15:58:41 -07:00
Matthias Hanel
5b9d20871d Exposing reserved memory in jsz/varz 
Signed-off-by: Matthias Hanel <mh@synadia.com>
 2021-09-16 18:55:50 -04:00
Derek Collison
670a81e213 Merge pull request #2538 from nats-io/batch-expires 
Make large batch requests expire more efficiently.
 2021-09-16 15:19:17 -07:00
Derek Collison
40a4d40337 Make large batch requests expire more efficiently. 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-16 15:00:18 -07:00
Derek Collison
296a568918 Merge pull request #2537 from nats-io/consumer-info-race 
[FIXED] ConsumerInfo in clustered mode during setup returned NotFound.
 2021-09-16 12:53:03 -07:00
Derek Collison
d91a0061ed Fix for data race around c.echo 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-16 11:59:52 -07:00
Derek Collison
9918f16914 ConsumerInfo in clustered mode would return not found if assigned but not properly setup with a leader. 
This change returns an info in those states now to better support client logic when binding consumers.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-16 10:56:34 -07:00
Ivan Kozlovic
47359c453e Merge pull request #2533 from nats-io/js_cons_fc_hb 
[CHANGED] JetStream: flow control requires heartbeats
 2021-09-16 10:20:48 -06:00
Derek Collison
decbae30ae Merge pull request #2535 from nats-io/issue-2528 
Allow clustered JetStream to allow duplicate stream creation calls.
 2021-09-16 08:49:25 -07:00
Derek Collison
cfbc69b12c Allow clustered JetStream to allow duplicate stream creation like single server mode. 
Resolves #2528

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-15 20:18:44 -07:00
Ivan Kozlovic
2f579abf0a Merge pull request #2532 from nats-io/mqtt_add_stream_check 
MQTT add test to check for idempotent session stream creation.
 2021-09-15 10:37:23 -06:00
Ivan Kozlovic
ec774c9347 MQTT handle idempotent session stream creation. 
Creating a stream will become idempotent, so assuming that we
should try to transfer the old session streams only on success
will no longer work.

Added a test that checks that "stream" list is queried only once
which means transfer was attempted only once after the second
cluster restart and new connection.

Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-15 10:15:45 -06:00
Ivan Kozlovic
108d4060ff [CHANGED] JetStream: flow control requires heartbeats 
The server will now reject the creation of a push consumer with
flow control if no heartbeat is set.

Resolves #2520

Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-15 09:50:15 -06:00
Ivan Kozlovic
d6f9a06f05 Merge pull request #2530 from nats-io/mqtt_del_sess 
MQTT: delete the session record even on restart with clean flag
 2021-09-15 09:01:49 -06:00
R.I.Pienaar
eac12f8dd0 Merge pull request #2531 from ripienaar/err_10154 
fix error code for JSStreamHeaderExceedsMaximumErr
 2021-09-15 16:29:35 +02:00
R.I.Pienaar
6a5ac92eec fix error code for JSStreamHeaderExceedsMaximumErr 
Also checks for gaps in codes

Signed-off-by: R.I.Pienaar <rip@devco.net>
 2021-09-15 09:01:03 +02:00
Derek Collison
765d39a2b6 Fix flapping test 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-14 17:47:14 -07:00
Ivan Kozlovic
4cc2677573 MQTT: delete the session record even on restart with clean flag 
When a session record exists and is currently not connected, if
the user reconnects with the same client ID but with now the
clean flag set, we are required to clear the state. In earlier
implementation (where we were using a stream per session), we
were not deleting the stream to be created right away. Since now
we just have a message per session, it is ok to delete that
message when clearing the existing session that is going to be
replaced.

Also made apiDispatch execute in place for any connection that
is not ROUTER or GATEWAY.

Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-14 16:33:48 -06:00
Derek Collison
777e3236f1 Merge pull request #2529 from nats-io/ln-js-domains 
Fixed a condition where JetStream assets could be created in multiple leafnode domains.
 2021-09-14 15:25:20 -07:00
Derek Collison
ebc581012e Fixed a condition where JetStream assets could be created in multiple leafnodes. 
Also added in optional Domain to StreamInfo.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-14 14:17:49 -07:00
Derek Collison
8e4a1b541f Merge pull request #2318 from c0deaddict/feature/systemd-hardening 
Add hardened systemd service
 2021-09-14 13:23:40 -07:00
Derek Collison
20574ffaad Merge pull request #2526 from nats-io/issue-2525 
[FIXED] #2525
 2021-09-14 06:41:18 -07:00
Derek Collison
9cdab0682a Bump version 
Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-13 18:41:12 -07:00
Derek Collison
4cfffc6fa6 We use u16 to encide header len when replicating JetStream messages. 
Make sure to error if we exceed that limit.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-13 18:13:07 -07:00
Derek Collison
6e37b98c8a Merge pull request #2522 from nats-io/fs-compact-mqtt 
During compaction the cache may not be loaded completely
 2021-09-13 16:27:10 -07:00
Derek Collison
620b56e12f During compaction the cache may not be loaded completely if msg block was lmb (active writing). 
This could lead to the filtered subject state being incorrect.

Signed-off-by: Derek Collison <derek@nats.io>
 2021-09-13 14:36:50 -07:00
Ivan Kozlovic
6e705c2dec Merge pull request #2519 from nats-io/fix_2510 
[FIXED] Websocket: possible panic when decoding CLOSE frame
 2021-09-13 14:45:20 -06:00
Ivan Kozlovic
2495180d76 Merge pull request #2518 from nats-io/fail_auth_token_where_not_supported 
[FIXED] Report config error if "token" defined for cluster{} and gateway{}
 2021-09-13 14:23:25 -06:00
Ivan Kozlovic
8d41daf6ab Updates based on code review 
Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-13 14:06:17 -06:00
Ivan Kozlovic
5c5cc88c1f [FIXED] Report config error if "token" defined for cluster{} and gateway{} 
Authentication token has never been supported for cluster and
gateway (and leafnode). There is not even a Token option in
ClusterOpts or GatewayOpts.

However, the parsing of the configuration was not rejecting this
misconfiguration, making users believe that token was used for
authentication.

Documentation will also be fixed since it is reported there that
token is supported, which again, has never been the case.

Signed-off-by: Ivan Kozlovic <ivan@synadia.com>
 2021-09-13 10:21:00 -06:00