Commit Graph

19711 Commits

Author SHA1 Message Date
Shawn
8ebd233a46 chore: update influxdb-erl to 1.1.12 2024-01-04 10:20:52 +08:00
zhongwencool
90404b62a4 fix: incorrectly attempt to update the field value when the previous field don't have value 2024-01-04 10:12:34 +08:00
JianBo He
b4c4103eab
Merge pull request #12246 from id/0102-docker-do-not-expose-11883
chore: do not expose 11883 by default in docker as we do not listen on it
2024-01-04 09:14:03 +08:00
Shawn
19e2ec9748 fix: update influxdb testcases 2024-01-03 18:39:29 +08:00
zhongwencool
f2f099ff18
Merge pull request #12249 from zhongwencool/log-prompt-garbled
fix: the API prompt garbled characters for updating the configuration
2024-01-03 16:00:46 +08:00
zhongwencool
00e8a36dbe fix: the API prompt garbled characters for updating the configuration 2024-01-03 15:34:45 +08:00
JianBo He
57c0a7c5ef
Merge pull request #12235 from zhongwencool/ft-config-update-api
fix: ft config update api return raw config
2024-01-03 14:07:22 +08:00
Ivan Dyachkov
b33f46f82a
Merge pull request #12248 from id/0102-prep-5.4.1-alpha.1
5.4.1-alpha.1
2024-01-02 15:21:30 +01:00
Ivan Dyachkov
d4c83cf463 chore: tag 5.4.1-alpha.1 2024-01-02 14:37:45 +01:00
Shawn
904bd0270f chore: bump app versions 2024-01-02 18:04:06 +08:00
Shawn
7c7a0b2172 refactor: split influxdb bridges to actions and connectors 2024-01-02 18:01:22 +08:00
Ivan Dyachkov
4c761b0ff9 docs: add changelog entry 2024-01-02 10:50:15 +01:00
Ivan Dyachkov
1649f67996 chore: do not expose 11883 by default in docker as we do not listen on it 2024-01-02 10:07:05 +01:00
Ivan Dyachkov
071d15d13e
Merge pull request #12228 from clementnuss/cjmn/ipv6-port-fix
feat(helm): only keep last digits for ipv6 ports
2024-01-02 08:41:43 +01:00
Clément Nussbaumer
c22687e322
feat(helm): only keep last digits for ipv6 ports
permits specifying ports such as :::18083, a somewhat strange but needed
syntax when trying to make emqx listen on IPv6 addresses
The solution here is to split by colon `:` and to only keep the last
entry, which will corespond to the actual number for the port.

Signed-off-by: Clément Nussbaumer <clement@n8r.ch>
2024-01-01 16:14:48 +01:00
Andrew Mayorov
a87df28dfc
Merge pull request #12243 from keynslug/fix/EMQX-11483/route-async-repl
fix(router): rely on local sync state updates to propagate routes
2023-12-29 16:31:00 +01:00
Andrew Mayorov
42da329b34
chore: add changelog entry 2023-12-29 14:05:50 +01:00
Andrew Mayorov
5c91984ad7
chore: leave comment describing need for tighter synchronization 2023-12-29 14:04:48 +01:00
Ilya Averyanov
02110bc79d
Merge pull request #12245 from savonarola/1229-improve-emqx_ds-interface
feat(ds): restore original add_generation/update_db_config callback semantics
2023-12-29 14:40:42 +02:00
Ilya Averyanov
9e0f3ce53b feat(ds): restore original add_generation/update_db_config callback semantics 2023-12-29 13:12:15 +03:00
JimMoen
a3fc4fa5c3
Merge pull request #12242 from JimMoen/1228-chore-typo-and-style
chore: typo and style
2023-12-29 16:39:03 +08:00
JimMoen
8f8ad2895b
chore: bump git-blame-ignore-revs 2023-12-29 09:08:21 +08:00
JimMoen
5e100f52b8
style: erlfmt all rebar.config files and bin/nodetool 2023-12-29 09:08:03 +08:00
JimMoen
0f10d9a8bb
build: erlfmt rebar.config files and bin/nodetool 2023-12-29 09:08:03 +08:00
JimMoen
a3f33cc8bc
chore: fix typos in code and suite 2023-12-29 09:07:51 +08:00
JimMoen
a35ed4dacb
docs: enhance markdown indent and fix typos 2023-12-29 09:07:26 +08:00
JimMoen
572f465346
fix(typo): sync to enterprise
- follow up emqx/emqx#10398
2023-12-29 09:07:26 +08:00
JimMoen
ed10827343
docs: README spell 2023-12-29 09:07:25 +08:00
Andrew Mayorov
b9627c420f
fix(router): rely on local sync state updates to propagate routes
Instead of (potentially lagging) global table state itself. Local
state should be the only source of truth anyway. Moreover,
synchronize parts of local state update logic through the broker
pool to ensure consistency.

The problem was that `emqx_router:has_route/2` check may observe a
stale route, which deletion is not yet replicated from the core node
to the local replicant node.

For example:
1. The only one subscriber per a given topic A unsubscribes from
   a replicant node.
2. The route to A is deleted by the emqx_broker.
3. Mria makes RPC to a core node, it succeeds and returns.
4. The client resubscribes or another client subscribes to the same
   topic A.
5. emqx_broker tries to add a route again:
   `emqx_router:do_add_route(Topic)`.
6. `emqx_router` checks if the route is present.
7. The stale route is present because deletion not replicated yet,
   so no route is being added.
8. Route deletion is replicated locally but it's too late:
   we already have a local subscriber without a route.

Co-Authored-By: Serge Tupchii <serge.tupchii@protonmail.com>
2023-12-28 19:16:31 +01:00
Andrew Mayorov
331bfaa535
fix(broker): avoid reclaiming per-topic counter twice 2023-12-28 17:43:50 +01:00
Andrew Mayorov
a8cd609ff4
chore(broker): simplify subscribe code path 2023-12-28 17:43:42 +01:00
Zaiming (Stone) Shi
2b61f5290d
Merge pull request #12236 from zmstone/1227-ensure-short-mqtt-bridge-clientid
fix(bridge/mqtt): ensure short clientid
2023-12-28 16:06:09 +01:00
zhongwencool
a6acebb4aa
Merge pull request #12240 from zhongwencool/ft-api-e-541
fix: return raw config when update ft's config
2023-12-28 22:25:22 +08:00
Andrew Mayorov
0eb737c96e
Merge pull request #12241 from keynslug/fix/EMQX-11665/s3-headers
fix(s3): handle extra configured headers correctly
2023-12-28 15:23:53 +01:00
Andrew Mayorov
72677f69d4
chore: add changelog entry 2023-12-28 14:49:05 +01:00
Andrew Mayorov
18dc9b2516
fix(s3): remove emqx_bridge_http from runtime dependencies
Otherwise it's impossible to test `emqx_s3` in isolation.
2023-12-28 14:13:22 +01:00
Andrew Mayorov
1d9374125c
chore: bump emqx_s3 to 5.0.13 2023-12-28 14:12:33 +01:00
Andrew Mayorov
33ef964582
fix(s3): handle extra configured headers correctly
* Header names in config become atoms.
  (Probably because of blanket `unsafe_atom_key_map/1` in `emqx_config`)
* Header names should be normalized to lowercase.
  (Otherwise signature will be incorrect)
2023-12-28 13:55:05 +01:00
Zaiming (Stone) Shi
4669c4d552 fix(bridge/mqtt): ensure short clientid
Some mqtt brokers do not allow long client IDs.
To make it compatible with this limitation, this commit
tries to limit the number of bytes under 23 with a best-effort
attempt to derive it from the bridge name.
2023-12-28 11:53:27 +01:00
JianBo He
d26f56e99b
Merge pull request #12238 from emqx/compatible-bridge-http-conf
fix(bridge-http): compatible conf with the error format introduced in 5.3.2
2023-12-28 16:18:44 +08:00
zhongwencool
1fd67564ce fix: return raw config when update ft's config 2023-12-28 14:49:36 +08:00
JianBo He
3ce2f225d7 test: update the test config 2023-12-28 13:48:51 +08:00
JianBo He
c0aba14d27 chore: update changes 2023-12-28 11:41:20 +08:00
JianBo He
b6391cf18c fix(bridge-http): compatible conf with the error format introduced in 5.3.2 2023-12-28 11:36:48 +08:00
SergeTupchiy
2d209ec576
Merge pull request #12234 from SergeTupchiy/EMQX-11643-fix-otel-conf-incompatibility
fix(emqx_opentelemetry): use converter to convert legacy metrics config
2023-12-27 15:44:28 +02:00
zhongwencool
e645fa4874 fix: ft config update api return raw config 2023-12-27 21:13:47 +08:00
Serge Tupchii
423b8b4274 fix(emqx_opentelemetry): use converter to convert legacy metrics config
HOCON converter is cleaner and safer option,
that will also work for configurations defined in emqx.conf.
Previously used `upgrade_raw_conf/1` is currently not supported in hocon_cli (used in bin/emqx).

Fixes: EMQX-11643
2023-12-27 14:15:12 +02:00
zhongwencool
711a9a06c8
Merge pull request #12221 from thalesmg/test-flaky-retry-m-20231222
test: reduce flakiness
2023-12-27 14:30:25 +08:00
JianBo He
4ebf776f65
Merge pull request #12232 from zhongwencool/force-leave-clean-rpc-commit
fix: remove rpc commit info when force_leave cluster
2023-12-27 09:25:17 +08:00
zhongwencool
11feab983f fix: clean rpc_commit when force_leave cluster 2023-12-26 15:25:58 +08:00