Charm stuck on JAAS

[marceloneppel]

Steps to reproduce

Deploy PostgreSQL K8S from stable in jimm.operatorinc.org

Expected behavior

Charm starts correctly.

Actual behavior

Charm stuck with the awaiting for cluster to start message.

Versions

Juju CLI: 3.1.6

Juju agent: 3.1.5

Charm revision: 158

Pebble logs:

root@pge-0:/# pebble logs
2023-11-01T09:31:09.897Z [postgresql] The files belonging to this database system will be owned by user "postgres".
2023-11-01T09:31:09.897Z [postgresql] This user must also own the server process.
2023-11-01T09:31:09.897Z [postgresql] 
2023-11-01T09:31:09.897Z [postgresql] The database cluster will be initialized with locales
2023-11-01T09:31:09.897Z [postgresql]   COLLATE:  C
2023-11-01T09:31:09.897Z [postgresql]   CTYPE:    C.UTF-8
2023-11-01T09:31:09.897Z [postgresql]   MESSAGES: C
2023-11-01T09:31:09.897Z [postgresql]   MONETARY: C
2023-11-01T09:31:09.897Z [postgresql]   NUMERIC:  C
2023-11-01T09:31:09.897Z [postgresql]   TIME:     C
2023-11-01T09:31:09.897Z [postgresql] The default text search configuration will be set to "english".
2023-11-01T09:31:09.897Z [postgresql] 
2023-11-01T09:31:09.897Z [postgresql] Data page checksums are enabled.
2023-11-01T09:31:09.897Z [postgresql] 
2023-11-01T09:31:09.897Z [postgresql] creating directory /var/lib/postgresql/data/pgdata ... ok
2023-11-01T09:31:09.898Z [postgresql] creating subdirectories ... ok
2023-11-01T09:31:09.899Z [postgresql] selecting dynamic shared memory implementation ... posix
2023-11-01T09:31:09.899Z [postgresql] selecting default max_connections ... 20
2023-11-01T09:31:10.739Z [postgresql] selecting default shared_buffers ... 400kB
2023-11-01T09:31:14.325Z [postgresql] selecting default time zone ... Etc/UTC
2023-11-01T09:31:14.344Z [postgresql] creating configuration files ... ok
2023-11-01T09:31:14.345Z [postgresql] running bootstrap script ... Bus error (core dumped)
2023-11-01T09:31:14.509Z [postgresql] child process exited with exit code 135
2023-11-01T09:31:14.509Z [postgresql] initdb: removing data directory "/var/lib/postgresql/data/pgdata"
2023-11-01T09:31:14.511Z [postgresql] pg_ctl: database system initialization failed
2023-11-01T09:31:15.084Z [postgresql] Traceback (most recent call last):
2023-11-01T09:31:15.084Z [postgresql]   File "/usr/bin/patroni", line 33, in <module>
2023-11-01T09:31:15.084Z [postgresql]     sys.exit(load_entry_point('patroni==3.0.2', 'console_scripts', 'patroni')())
2023-11-01T09:31:15.084Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/__main__.py", line 144, in main
2023-11-01T09:31:15.084Z [postgresql]     return patroni_main()
2023-11-01T09:31:15.085Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/__main__.py", line 136, in patroni_main
2023-11-01T09:31:15.085Z [postgresql]     abstract_main(Patroni, schema)
2023-11-01T09:31:15.085Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/daemon.py", line 181, in abstract_main
2023-11-01T09:31:15.085Z [postgresql]     controller.run()
2023-11-01T09:31:15.085Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/__main__.py", line 106, in run
2023-11-01T09:31:15.085Z [postgresql]     super(Patroni, self).run()
2023-11-01T09:31:15.086Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/daemon.py", line 126, in run
2023-11-01T09:31:15.086Z [postgresql]     self._run_cycle()
2023-11-01T09:31:15.086Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/__main__.py", line 109, in _run_cycle
2023-11-01T09:31:15.086Z [postgresql]     logger.info(self.ha.run_cycle())
2023-11-01T09:31:15.086Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/ha.py", line 1770, in run_cycle
2023-11-01T09:31:15.086Z [postgresql]     info = self._run_cycle()
2023-11-01T09:31:15.087Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/ha.py", line 1592, in _run_cycle
2023-11-01T09:31:15.087Z [postgresql]     return self.post_bootstrap()
2023-11-01T09:31:15.087Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/ha.py", line 1483, in post_bootstrap
2023-11-01T09:31:15.087Z [postgresql]     self.cancel_initialization()
2023-11-01T09:31:15.087Z [postgresql]   File "/usr/lib/python3/dist-packages/patroni/ha.py", line 1476, in cancel_initialization
2023-11-01T09:31:15.088Z [postgresql]     raise PatroniFatalException('Failed to bootstrap cluster')
2023-11-01T09:31:15.088Z [postgresql] patroni.exceptions.PatroniFatalException: 'Failed to bootstrap cluster'

Additional context

Huge pages setting in the unit:

root@pge-0:/# sysctl -a | grep hugepage
sysctl: reading key "kernel.apparmor_display_secid_mode"
sysctl: reading key "kernel.unprivileged_userns_apparmor_policy"
vm.nr_hugepages = 1024
vm.nr_hugepages_mempolicy = 1024
vm.nr_overcommit_hugepages = 0
root@pge-0:/#

MM discussion

[github-actions]

https://warthogs.atlassian.net/browse/DPE-2880

[taurus-forever]

From https://chat.charmhub.io/charmhub/pl/yg6asxao6f8hpn93xpq83efrte Marcelo wrote:

I created a new revision for the PostgreSQL charm on the 14/edge/test channel. It should have the fix to start PostgreSQL (and avoid it being stuck with the awaiting for cluster to start message). I couldn't reproduce the Juju secrets issue that you showed me in the logs, but now the PostgreSQL charm should be able to start.
....
If that revision works, I should create a pull request to later publish the revision correctly to the 14/edge channel.

We are waiting for the fix confirmation to merge in edge...

[taurus-forever]

@marceloneppel what is our plan here?

[marceloneppel]

We'll need some help from Alex Kilroy to bootstrap the environment again. He was busy, so that we couldn't progress in this task.

[taurus-forever]

Hi @ale8k , is it still reproducible on JAAS (with 14/stable or 14/candidate we are preparing for stable release now)?
If so, can you help us with env to reproduce and troubleshoot there? Tnx!

[taurus-forever]

Dear @ale8k are there any place we can reproduce this issue (see my comment above)? Tnx!

[ale8k]

Hi @taurus-forever, I'm unsure on how to reproduce this... Perhaps @kian99 knows?

[ale8k]

Ahh! This can be ignored, operator day has passed.

[taurus-forever]

Resolving the ticket as no longer topical.

Data Team is still interested in JAAS deployment/testing and searching the environment to test and document it (separate story).

Tnx!