ISCSI lun accesed by two nodes

V

virven

New Member
Feb 27, 2018
4
0
1
43
Hi, I have setted up a Proxmox cluster of 2 nodes :

at first I've added an iscsi target to my first node - my newly created proxmox hypervisor, than added lvm and setted up vms on that lun - all worked well untill Ive added a second node to cluster( another proxmox hypervisor) to get cluster of 2 nodes.

When two nodes are tring to acces the very same iscsi target (in one datacenter ) one of them logs in perfectly via created portal with credentials and such but the other one gets kicked out with no available portal msg.
I've added another iqn for the other proxmox host, I have created specyfic acls and added another portal for the other host but stil no change. I am unable to use iscsi lun from both proxmox servers... just from the first one.

any ideas ?
 
Hi LnxBil, thank You for that ...
my iscsi runs @ debian 9 via targetcli I'm able to add even more than 2 initiators to acces my ISCSI target but I don't understand how 2 or more at the same time can join to one space ..
Do both proxmox hypervisors use the very same iqn pass and login ?
AFAIK iqn for any of Proxmox nods should be uniqe so thay can't share the same credentials when pointing to one lun, right ?
What I was trying to do was to add in luns acls one iqn for he 1st proxmox node (ofc login and pass) and another acls with iqn for the second one.
I have even created more than one portal so both of them can join on different port - I've ended up with Proxmox primary node working with ISCSI LUN and Proxmox 2nd node acting like working with it ... what I mean is that it looks like being online but I'm unable to brows and I get timeout trying doing that.

When setting up iscsi I've used server-world.info tutorial,which I'm unable link :)
 
I'm using Debian 8 iSCSI (they switched the iSCSI backend in 9+) and I only create a LUN (two entries in /etc/iet/ietd.conf) and login from all nodes and can use it without any problems. Maybe it's something new with the Debian 9 iSCSI implementation.
 
okay, thanks for that mate, I'll track this issue, can You please post those two entries You're talking about ... might give me some tips , if You don't mind ofc


EDIT:
by now I'm stuck with second node :
iscsiadm: default: 1 session requested, but 1 already present.

seems like iscsi target allows just 1 session for a lun and I have no idea where to set more than 1 by targetcli

anyway seems like Your tip about multiple logins is the key to my problem
 
Last edited:
Normally I don't Reply to my post cuz it's lame but this time I'll do so ...

Mark as [PROBLEM SOLVED]
how ?
by #pvecm updatecerts

I have no knowladge why certyficates failed to set while joinin both node's togeather but still one had missmatch , and as far as I can understad replication failed due to certyficates problem . When certs matched, replication started and I have ISCSI target on both nodes.
I'm not sure how it works, some sort of pipe is created due to replication process, but what I can say is that one session works like a charm for 2 nodes, probably another node's will do same way.
Now I can live migrate VM's like I should.

Anyway thanks a lot for all that help and joining my battle with ISCSI problems.

Best regards.
Virven
 
Last edited:
Good that you solved your problem!

Only for documentation if someone stumbles over this thread:
In the meantime I setup a Debian 9 iSCSI server with istgt for another project and it is similar easy to setup as ietd, so both are working fine with Debian. With the stock-configuration of both, you only need to add the LUN you want to export in the configuration and then you're good to go.
 
You must log in or register to reply here.
Top Bottom