Dell iSCSI issues

N

nick-scs

New Member
Feb 13, 2024
2
0
1
I am configuring a PVE (8.1.4) cluster on 3 Dell R640s with iSCSI from a Dell MD3620i.
Each server has 2 NICs for iSCSI, each on a different subnet and switch and the each of the MD3620s controllers had an address on those subnets respectively.
When I add the storage to PVE all hosts start giving errors and eventually become unresponsive. I can recover them by removing the disk from storage.cfg.

Screenshot 2024-02-13 at 15.50.11.png

I can get the iSCSI working on a single host with a single connection but as soon as I add multiple paths it fails. I have configured multipath-tools as per the guide and have almost identical equipment working elsewhere.





Very grateful for any advice!
 
This seems to be a network problem. The messages are pretty unambiguous that a 5 second timeout has expired. Thats a long time to not receive a reply to a health check message. I would double check your MTUs, traceroutes, network, subnets, duplicate IPs, etc.
The timeout is on a iSCSI layer, below multipath.

You can also reach out to your storage vendor, as the problem is at Kernel, iSCSI, TCP layer - before it gets to PVE. Ask them for troubleshooting steps for iSCSI connectivity to Debian Linux.

Good luck

Blockbridge: Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox
 
Thanks @bbgeek17
I wish I could say it was the first time I've lost a few hours over a single MTU setting. Appreciate the sense check. All working excellently now. :)
 
You must log in or register to reply here.
Top Bottom