Ceph » mgr » Dashboard

I just upgraded to 18.2.0. The new dashboard home screen is a regrettable step backward. Do something constructive with all the wasted space on the top status tile. That is not the topic of this issue, so forgive me for saying it, but the shock is still rather fresh. The objects and PG Status gauges on the previous dashboard were very useful.

My issue is that this is the third consecutive major release of Ceph that has broken my RGW dashboard for completely stupid and avoidable reasons.

The upgrade itself went reasonably well, and upon completion I start testing the components and start getting a 500 error on the Object Gateway tab of the mgr dashboard.

The mgr log is giving throwing this when I try to make a request on a RGW page:

bash[32475]: Internal Server Error
bash[32475]: Traceback (most recent call last):
bash[32475]:   File "/usr/share/ceph/mgr/dashboard/services/exception.py", line 47, in dashboard_exception_handler
bash[32475]:     return handler(*args, **kwargs)
bash[32475]:   File "/lib/python3.6/site-packages/cherrypy/_cpdispatch.py", line 54, in __call__
bash[32475]:     return self.callable(*self.args, **self.kwargs)
bash[32475]:   File "/usr/share/ceph/mgr/dashboard/controllers/_base_controller.py", line 263, in inner
bash[32475]:     ret = func(*args, **kwargs)
bash[32475]:   File "/usr/share/ceph/mgr/dashboard/controllers/_rest_controller.py", line 191, in wrapper
bash[32475]:     return func(*vpath, **params)
bash[32475]:   File "/usr/share/ceph/mgr/dashboard/controllers/rgw.py", line 108, in list
bash[32475]:     'port': int(metadata['frontend_config#0'].split('port=')[1])
bash[32475]: ValueError: invalid literal for int() with base 10: '443 ssl_certificate=config://rgw/cert/rgw.NBDFSv2.NBDFS.C3'
bash[32475]: [::ffff:10.3.34.90:32990] [GET] [500] [0.004s] [nbadmin] [513.0B] /api/rgw/daemon
bash[32475]: [b'{"status": "500 Internal Server Error", "detail": "The server encountered an unexpected condition which prevented it from fulfilling the request.", "request_id": "ba61646e-eb89-43eb-89d0-156ec3ffc2b7"}

For AGES, since mimic or perhaps earlier, my rgw_frontends looked like this:

beast port=80 ssl_port=443 ssl_certificate=config:/blah/blah/blah

And the various mgrs have been able to parse the port numbers no problem. Now in reef you are splitting on 'port='? It never occurred to anyone that more config data might come afterward? You don't even want to do a second split on a non-numeric char on the first split result just in case?

Not everyone puts the port at the end. If that is now the requirement then put it in the documentation along with an example of a valid rgw_frontends!

You have a big key value store so I don't know why rgw_frontends has to try to jam everything into 1 key in the first place, creating opportunities for mistakes like this.