PBS 2.2-3 lose management connection (on several PBS, no problems on 2.2-1)

[dea]

I have an intermittent problem with several PBS upgraded to the 2.2-3 release (previously the 2.2-1 version had no problems). The problems occur (identical) both on a PBS installed in bare metal (the one in the screenshot) and on a PBS installed in a Proxmox VM.

Every now and then the connection with the PBS is lost, but all the processes (backup on tape, access to the datastore via NFS on a dedicated point-to-point network, checks, etc. etc.) do not present the slightest problem.

The summary is this:



Repeat, with the 2.2-1 release everything was perfect. Some idea ?

 

[dcsapak]

does the journal/syslog say anything of interest? what does the browser console say (you can open that with 'F12' in most browsers) ?

 

[dea]

does the journal/syslog say anything of interest? what does the browser console say (you can open that with 'F12' in most browsers) ?

I can see.. only an additional note... after 10-15 seconds resume the functionality. This problem present on two different PBS, one bare metal and one virtual (on Proxmox), the common components are the datastore: NFS for both, but the datastore is connected to a specific network, different from the management and interface to Proxmox VE of the PBS.

 

[dea]

Another problem two minutes ago:

 

[dea]

Nothing interesting in the logs. On the VM, if I roll back to 2.2-1 the problem goes away.

 

[dcsapak]

mhmm.. thats strange, there isn't much that would affect the connections between those versions. could you post the journal during the connection fails? maybe we see something relevant there that you missed?

 

[dea]

mhmm.. thats strange, there isn't much that would affect the connections between those versions. could you post the journal during the connection fails? maybe we see something relevant there that you missed?

It is not a problem with the network connection itself. When the problem occurs, I tried to open a new connection to the login page and it doesn't respond (the login page), but if I connect in ssh to the server there is no problem. After about 15 seconds it resumes and everything works.

 

[dea]

The problem still occurred several times. The GUI times out, but the background functions continue their work. In addition to the GUI, the problem occurs on Proxmox VE which is unable to access (for a few tens of seconds) the PBS storage, while functions such as verification, sync, etc. they do not present obvious problems (nor errors in the various system logs). The problematic PBS running on a VM was reset to 2.2-1 and the problems disappeared. Can I roll back with the bare metal version? Both systems use NFS datastores, but there are no anomalies in the system logs.