Fixes https://emqx.atlassian.net/browse/EMQX-10944
Also updates ekka -> 0.15.15, mria -> 0.6.4
How to test
===========
1. Start 2 or more EMQX nodes and merge them in a cluster.
2. Stop them in order.
3. Start only the first node that was stopped in the previous step.
4. Wait until the log is printed.
Or, more easily:
1. Start 2 or more EMQX nodes and merge them in a cluster.
2. Stop all but one.
3. Run `mria_mnesia:diagnosis([]).` on that node.
Example output
==============
```
Check check_open_ports should get ok but got #{msg =>
"some ports are unreachable",
results =>
#{'emqx@172.100.239.4' =>
#{open_ports =>
#{4370 => false,
5370 =>
false},
ports_to_check =>
[4370,5370],
resolved_ips =>
[{172,100,239,
4}],
status =>
bad_ports},
'emqx@172.100.239.5' =>
#{open_ports =>
#{4370 => false,
5370 =>
false},
ports_to_check =>
[4370,5370],
resolved_ips =>
[{172,100,239,
5}],
status =>
bad_ports}}}
```
After one node is back:
```
Check check_open_ports should get ok but got #{msg =>
"some ports are unreachable",
results =>
#{'emqx@172.100.239.4' =>
#{ports_to_check =>
[4370,5370],
resolved_ips =>
[{172,100,239,
4}],
status => ok},
'emqx@172.100.239.5' =>
#{open_ports =>
#{4370 => false,
5370 =>
false},
ports_to_check =>
[4370,5370],
resolved_ips =>
[{172,100,239,
5}],
status =>
bad_ports}}}
```
This fixes a bug in the protobuf schema registry functionality. Before
this fix one would get a badarith error if one tried to assign a float
value to an uint64 field. However, this commit fixes this by upgrading
gpb so we instead will get a gpb_type_error which is what we want.
Fixes:
https://emqx.atlassian.net/browse/EMQX-10775
Previously, the div operation could only be used as an infix operation
while mod could only be used as a function call. After this commit, one
can use both div and mod using function call syntax and infix syntax.
Fixes: https://emqx.atlassian.net/browse/EMQX-10216
* release-50:
fix(pulsar): use a binary duration as default `health_check_interval`
docs: add changelog entry
docs: clarify description of bridge username and password
chore: bump to v5.0.25
fix(limiter): adjust type for compatibility
fix(limiter): fix that update node-level limiter config will not working
chore: upgrade dashboard to v1.2.4-1 for ce
chore: upgarde rulesql to 0.1.6 to fix invaid utf8 input
chore: add changelog for 10659
fix: crash when sysmon.os.mem_check_interval = disabled
chore: bump influxdb version && update changes
refactor(influxdb): move influxdb bridge into its own app
chore: add listener default changelog
fix: ocsp cache SUITE failed
fix: ensure atom key for emqx_config:get
fix: only fill cerf_file default in server side
fix: authn init is empty
fix: bad listeners default ssl_options
Includes fix: https://github.com/uwiger/gproc/pull/193
Prior to the fix, when using the `random` pool strategy, one of the
workers receives about double the load of other workers, which
decreases throughput of bridges like webhook.
* master: (194 commits)
fix(limiter): update change && fix deprecated version
chore: update changes
perf(limiter): simplify the memory represent of limiter configuration
ci(perf test): update tf variable name and set job timeout
ci: fix artifact name in scheduled packages workflow
fix: build_packages_cron.yaml workflow
ci: move scheduled builds to a separate workflow
build: check mnesia compatibility when generating mria config
docs: fix a typo in api doc description
feat(./dev): use command style and added 'ctl' command
test: fix delayed-pubish test case flakyness
refactor: remove raw_with_default config load option
chore: add changelog for trace timestrap
feat: increase the time precision of trace logs to microseconds
chore: make sure topic_metrics/rewrite's default is []
docs: Update changes/ce/perf-10417.en.md
chore: bump `snabbkaffe` to 1.0.8
ci: run static checks in separate jobs
chore(schema): mark deprecated quic listener fields ?IMPORTANCE_HIDDEN
chore: remove unused mqtt cap 'subscription_identifiers'
...
Fixes https://emqx.atlassian.net/browse/EMQX-9656
See also https://github.com/emqx/ehttpc/pull/45
This fixes a race condition where the remote server would close the
connection before or during requests, and, depending on timing, an
`{error, normal}` response would be returned. In those cases, we
should just retry the request without using up "retry credits".
ekka 0.15.1 uses mria 0.5.2, which includes the following changes:
- fix(mria_membership): call `mria_rlog:role/1` safely
- feat: add extra field to ?rlog_sync table (for future use)
Ivan Dyachkov