VSAN issues seemingly stemming from update to 1.6.580.7361

[CombFiltered]

I recently updated to 1.6.580.7361. I've had two main issues after this, I'm not sure if related or not. But I did not have the issues before, and now I do.

One minor but nuisance issue is that the Storage, Files shares page of the web interface will commonly be blank, saying there are no file shares. If I leave it on that page for a while, they will eventually show up, but then disappear again after some time.

Another more problematic issue is that a Proxmox host can no longer connect to its iSCSI target. It initially connects, then it looks like is disconnected by VSAN with the below errors.

5/5 10:57:52.224670 33 IMG: *** ImageFile_ScsiExec: SCSIOP (0xA3) is not supported.
5/5 10:57:52.823340 e9 Srv: *** SwSocket::Recv: Swn_SocketRecv() failed with error 10053 (0x2745)!
5/5 10:57:52.823387 e9 C[5], LIN: iScsiConnection::recvData: Recv returned 10053 (0x2745)!
5/5 10:57:52.823402 e9 C[5], LIN: iScsiConnection::receive: recvData returned error 10053 (0x2745)!
5/5 10:57:52.823434 e9 C[5], LIN: iScsiConnection::recvWorker: *** 'recv' thread: recv failed 10058.
5/5 10:57:52.823936 ea Tgt: iScsiTarget::closeSession: close 'iqn[redacted for privacy/security]': 1 session(s) opened, 65535 more allowed.

A bit after this, the sessions go to 0. Since it's also 0 before booting this machine, I can only assume that one session is from mounting for boot, and then the other from Proxmox actually mounting the file system?

I could probably dig through logs to find what the previous version was, but I would have no way to roll back that I know of.

[CombFiltered]

Looks like I was on 1.6.580.7359.

[yaroslav (staff)]

Can I please get the logs from your systems?

[CombFiltered]

Ticket 1318119 submitted with the logs. I forgot to mention this in the ticket specifically, but the logs were from 10 minutes after I had the iSCSI boot issue, so it should be very recent in all the logs.

[CombFiltered]

Ok I fully realize that a lot of the variables here are different, but I wonder if this is the same thing that happened here with VSAN and Proxmox. I had restarted the VSAN server trying to figure out what was going on with the dashboard and file shares disappearing. I wonder if that triggered a SCSI ID change on VSAN, and Proxmox doesn't like that? I don't know nearly enough to do a deep dive on this, but I saw enough similarities in the situations that I thought it might be worth looking into and making sure you at least knew of similar issues.

https://xcp-ng.org/forum/topic/891/iscs ... i-opcode/3
https://github.com/xcp-ng/xcp/issues/128
https://smcleod.net/2015/12/iscsi-scsi- ... rsistence/

[yaroslav (staff)]

Thanks for sharing! I have forwarded the logs and the original request to my colleagues.