[ceph-users] Re: ceph dashboard reef 18.2.2 radosgw
Pierre, This is indeed the problem. I modified the line in /usr/share/ceph/mgr/dashboard/controllers/rgw.py 'port': int(re.findall(r'port=(\d+)', metadata['frontend_config#0'])[0]) to just be: port: 443 and all works. I see that in the pull that if it cannot find port= or in one of endpoint or ssl_endpoint, it gets set to None. So even thoughrgw itself will listen on port 443 if I use ssl_endpoint without a port, I will probably have to explicitly set it for the dashboard to work. -Chris On Sunday, May 12, 2024 at 10:03:59 PM MDT, Pierre Riteau wrote: Hi Christopher, I think your issue may be fixed by https://github.com/ceph/ceph/pull/54764, which should be included in the next Reef release. In the meantime, you should be able to update your RGW configuration to include port=80. You will need to restart every RGW daemon so that all the metadata is updated. Best wishes, Pierre Riteau On Wed, 8 May 2024 at 19:51, Christopher Durham wrote: > Hello, > I am uisng 18.2.2 on Rocky 8 Linux. > > I am getting http error 500 whe trying to hit the ceph dashboard on reef > 18.2.2 when trying to look at any of the radosgw pages. > I tracked this down to /usr/share/ceph/mgr/dashboard/controllers/rgw.py > It appears to parse the metadata for a given radosgw server improperly. In > my varoous rgw ceph.conf entries, I have: > rgw frontends = beast ssl_endpoint=0.0.0.0 > ssl_certificate=/path/to/pem_with_cert_and_key > but, rgw.py pulls the metadata for each server, and it is looking for > 'port=' in the metadata for each server. When it doesn't find it based on > line 147 in rgw.py, the ceph-mgr logs throwan exception which the manager > proper catches and returns a 500. > Would changing my frontends definition work? Is this known? I have had the > frontends definition for awhile prior to my reef upgrade. Thanks > -Chris > ___ > ceph-users mailing list -- ceph-users@ceph.io > To unsubscribe send an email to ceph-users-le...@ceph.io > ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Re: ceph dashboard reef 18.2.2 radosgw
Hi Christopher, I think your issue may be fixed by https://github.com/ceph/ceph/pull/54764, which should be included in the next Reef release. In the meantime, you should be able to update your RGW configuration to include port=80. You will need to restart every RGW daemon so that all the metadata is updated. Best wishes, Pierre Riteau On Wed, 8 May 2024 at 19:51, Christopher Durham wrote: > Hello, > I am uisng 18.2.2 on Rocky 8 Linux. > > I am getting http error 500 whe trying to hit the ceph dashboard on reef > 18.2.2 when trying to look at any of the radosgw pages. > I tracked this down to /usr/share/ceph/mgr/dashboard/controllers/rgw.py > It appears to parse the metadata for a given radosgw server improperly. In > my varoous rgw ceph.conf entries, I have: > rgw frontends = beast ssl_endpoint=0.0.0.0 > ssl_certificate=/path/to/pem_with_cert_and_key > but, rgw.py pulls the metadata for each server, and it is looking for > 'port=' in the metadata for each server. When it doesn't find it based on > line 147 in rgw.py, the ceph-mgr logs throwan exception which the manager > proper catches and returns a 500. > Would changing my frontends definition work? Is this known? I have had the > frontends definition for awhile prior to my reef upgrade. Thanks > -Chris > ___ > ceph-users mailing list -- ceph-users@ceph.io > To unsubscribe send an email to ceph-users-le...@ceph.io > ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
[ceph-users] Re: ceph dashboard reef 18.2.2 radosgw
Hello Christopher, Could you please paste the logs and exceptions to this thread as well? Regards, Nizam On Wed, May 8, 2024 at 11:21 PM Christopher Durham wrote: > Hello, > I am uisng 18.2.2 on Rocky 8 Linux. > > I am getting http error 500 whe trying to hit the ceph dashboard on reef > 18.2.2 when trying to look at any of the radosgw pages. > I tracked this down to /usr/share/ceph/mgr/dashboard/controllers/rgw.py > It appears to parse the metadata for a given radosgw server improperly. In > my varoous rgw ceph.conf entries, I have: > rgw frontends = beast ssl_endpoint=0.0.0.0 > ssl_certificate=/path/to/pem_with_cert_and_key > but, rgw.py pulls the metadata for each server, and it is looking for > 'port=' in the metadata for each server. When it doesn't find it based on > line 147 in rgw.py, the ceph-mgr logs throwan exception which the manager > proper catches and returns a 500. > Would changing my frontends definition work? Is this known? I have had the > frontends definition for awhile prior to my reef upgrade. Thanks > -Chris > ___ > ceph-users mailing list -- ceph-users@ceph.io > To unsubscribe send an email to ceph-users-le...@ceph.io > > ___ ceph-users mailing list -- ceph-users@ceph.io To unsubscribe send an email to ceph-users-le...@ceph.io
