Problems after upgrade to PVE 8.1.4

[OlliT]

Hello.

sorry for my bad english.
After upgrade to PVE 8.1.4 I have one Problem but I think I have 2.

Cluster of 7 Nodes but only one has install the update and the follow problem.
The other Nodes are on PVE 8.0.4
I use Open-E Jovian as an Cluster ( Active/Passive )

Problems:
1. Mar 07 00:50:21 pve11 pvestatd[9760]: command '/usr/bin/iscsiadm --mode node --targetname iqn.b23.node-ha-01:cluster-01 --login' failed: exit code 15
2. Empty graphs

1. The Problem there is no timeout, the error keeps coming back. On the Nodes with 8.0.4 the error in Syslog stop after a few minutes.

I have read other threads but that did not help. I cannot disable IPv6 on Jovian ( Open-E ), there is no IPv6 in use i think ( on the GUI ).

Is there a solution or is there a way to install 8.0.4 again? There is no old ISO online. Or is 8.1.2 the solution?

 

[fiona]

Hi,

Problems:
1. Mar 07 00:50:21 pve11 pvestatd[9760]: command '/usr/bin/iscsiadm --mode node --targetname iqn.b23.node-ha-01:cluster-01 --login' failed: exit code 15
2. Empty graphs

The second issue is likely a consequence of the first, because the status daemon is responsible for both.

So man iscsiadm says the exit code means: 15 ISCSI_ERR_SESS_EXISTS - session is logged in.
What is the output of iscsiadm --mode node? Maybe a duplicate portal is defined?

Maybe related: https://bugzilla.proxmox.com/show_bug.cgi?id=5173

 

[OlliT]

Hi Fiona,

thanks for your replay.

Ok, so my fokus is on the first Problem.

The output from iscsiadm:

10.230.100.21:3260,1 iqn.2022-01.node-ha-01:cluster-01
10.230.100.22:3260,1 iqn.2022-01.node-ha-01:cluster-01
10.230.101.22:3260,1 iqn.2022-01.node-ha-01:cluster-01
10.230.102.22:3260,1 iqn.2022-01.node-ha-01:cluster-01
10.230.103.22:3260,1 iqn.2022-01.node-ha-01:cluster-01
10.230.100.21:3260,1 iqn.2022-10.node-ha-p-2:cluster-01
10.230.100.22:3260,1 iqn.2022-10.node-ha-p-2:cluster-01
10.230.101.22:3260,1 iqn.2022-10.node-ha-p-2:cluster-01
10.230.102.22:3260,1 iqn.2022-10.node-ha-p-2:cluster-01
10.230.103.22:3260,1 iqn.2022-10.node-ha-p-2:cluster-01

That is what I wondering for:

The IP 10.230.100.21 is the virtual IP Address for the first Target from the Cluster and
the IP 10.230.100.22 is the virtual IP Address for the second Target from the Cluster ( other Pool ).

10.230.101.22 is only the webgui from open-e ( not reachable, because other Networkcard and other Switch and no connection to the SAN Network )
10.230.102.22 is the physical IP Address from the active Cluster ( not reachable, because not configurard, because the right IP is the virtual IP )
10.230.103.22 is for the Synchronisation beetween the Cluster ( not Reachable because other Networkcard and no connection to SAN Network )

The bugzilla URL I have also read yesterday, but I found no solution for my Problem.

It is possible go back to the old config, this config works for the old PVE Version 8.0-4

Kind Regards
Oliver

 

[OlliT]

Hello,

any answer for this problem?

Kind Regards
Oliver

 

[fiona]

The issue seems to be that also non-targets are reported, so it's always attempted to re-login (to get a session for each target). Maybe you can configure that only actual targets are reported on the server side? Otherwise, you can try deleting those additional entries like suggested here:
https://bugzilla.proxmox.com/show_bug.cgi?id=5173#c17