Project

General

Profile

Actions

Bug #56745

closed

unrecognized option ssl_server_port appears to cause failure of MGR

Added by Brian Woods almost 2 years ago. Updated almost 2 years ago.

Status:
Duplicate
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
% Done:

0%

Source:
Tags:
Backport:
Regression:
No
Severity:
2 - major
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

This is going to take a second, sorry. This is a test environment and is open to destructive testing!

I deployed a fresh Ubuntu 20.04 system, deploy my defacto config scripts, docker and the basic Ceph Quincy 17.2.2 bootstrap to a VM, no OSDs. Then I deployed a second VM with two OSDs (one SSD and one HDD). Everything is working fine so far.

I then deployed a third install to a physical box with some drives and an SSD, monitoring services start on it and all is well. However, after deploying OSDs the the third node (second node with OSDs), withing sever seconds the manager service vanishes from the cluster.

Thought I did something wrong, wiped everything and re-deployed. Same exact issue...

I have been looking through the logs, and as best as I can tell, the MGR fails to deploy/redeploy due to these errors (private values/IPs removed):

cephadmin bash[2817]: debug   -228> 2022-07-27T23:49:30.438+0000 7fb0c162b700 10 set_mon_vals mgr/dashboard/server_addr = #.#.#.# (unrecognized option)
cephadmin bash[2817]: debug   -227> 2022-07-27T23:49:30.438+0000 7fb0c162b700 10 set_mon_vals mgr/dashboard/server_port = 8080 (unrecognized option)
cephadmin bash[2817]: debug   -226> 2022-07-27T23:49:30.438+0000 7fb0c162b700 10 set_mon_vals mgr/dashboard/ssl = false (unrecognized option)
cephadmin bash[2817]: debug   -225> 2022-07-27T23:49:30.438+0000 7fb0c162b700 10 set_mon_vals mgr/dashboard/ssl_server_port = 8443 (unrecognized option)
cephadmin bash[2817]: debug   -224> 2022-07-27T23:49:30.438+0000 7fb0c162b700 10 set_mon_vals mgr/orchestrator/orchestrator = cephadm (unrecognized option)
cephadmin bash[2817]: debug   -129> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/cephadm/container_init = True (unrecognized option)
cephadmin bash[2817]: debug   -128> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/cephadm/migration_current = 5 (unrecognized option)
cephadmin bash[2817]: debug   -127> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/dashboard/ALERTMANAGER_API_HOST = http://###########:9093 (unrecognized option)
cephadmin bash[2817]: debug   -126> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/dashboard/GRAFANA_API_SSL_VERIFY = false (unrecognized option)
cephadmin bash[2817]: debug   -125> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/dashboard/GRAFANA_API_URL = https://###########:3000 (unrecognized option)
cephadmin bash[2817]: debug   -124> 2022-07-27T23:49:30.450+0000 7fb09a6b2700 10 set_mon_vals mgr/dashboard/PROMETHEUS_API_HOST = http://###########:9095 (unrecognized option)

This causes the dashboard to attempt a starts up with a 0 for the port as best as I can tell:

from='client.? #.#.#.#:0/3815048734' entity='client.admin' cmd=[{"prefix": "mgr module enable", "module": "dashboard"}]: dispatch

Everything goes down hill from there, the mgr crashes as it can't talk to its own services, finally the docker image and host services are purged...

I am not sure if it is something in my (very bare bones) config somehow, or a bug.. But... Any help would be appreciated. I have not Ceph since Luminous, so I am very out of date to say the least...

Note, I am not tagging the second and third node with "_admin", as I thought that caused it the first time.


Related issues 1 (0 open1 closed)

Is duplicate of cephsqlite - Bug #55304: libcephsqlite: crash when compiled with gcc12 cause of regex treating '-' as a range operatorResolved

Actions
Actions

Also available in: Atom PDF