main exit

[baum]

main exit

1. do not catch System exit
2. return exit code 0 from TERM signal handler
3. tweak wait for gateways ha test

partial revert of

• 6c120cf
• 042cf8c

[VallariAg]

Teuthology run: https://pulpito.ceph.com/vallariag-2024-10-22_17:54:08-nvmeof-main-distro-default-smithi/

Explanation of 1 failed job:
During HA testing, ceph orch daemon rm nvmeof.nvmeof.a triggered cluster warning CEPHADM_DAEMON_PLACE_FAIL. It was cleared after the daemon restarted. I reran it and the test passed: https://pulpito.ceph.com/vallariag-2024-10-24_07:48:47-nvmeof-main-distro-default-smithi/

[caroav]

@VallariAg does it add a test to verify that if a gw exits with a bad status its being restarted? I guess no?

[VallariAg]

@caroav it includes HA testing (thrasher test) and that passed. Do you mean self-restart after thrashing? (if yes, then yeah, we are not checking for self-restart - let me know and I can add that)

[caroav]

this is a trashing case that we still need to add testing for in atom and teuthology. We need to kill one of {spdk, monitor client}, and see that the gw is exiting, but then being restarted automatically. Both cases should be checked (kill spdk, and in another iteration kill monitor client).

[VallariAg]

@caroav ah okay, understood, will talk to Barak about it too next week. thank you!