Start a Conversation

Solved!

Go to Solution

995

February 28th, 2023 07:00

ISCSI question on ME4024

Hello and thank you for you insights,

I have a PVE cluster and i want the volumes on ME4024 to be handled by multipathing on the cluster.

I configured my controllers like this :

 

Capture d’écran 2023-02-28 à 09.39.21.png

And configured 2 SAN switches with 2 separate non routed private vlan as suggested by Vendors.

On PVE i declared the 8 controllers ip as ISCSI Portals.

 

Capture d’écran 2023-02-28 à 16.31.43.png

 

And then when i map a volume on all my 3 initiators and use multipath i have theses errors :

iscsid: Kernel reported iSCSI connection 1:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 28 14:11:20 pve3 kernel: [ 543.556114] connection6:0: ping timeout of 5 secs expired, recv timeout 5, last rx 4295025408, last ping 4295026688, now 4295027968
Feb 28 14:11:20 pve3 kernel: [ 543.558146] connection6:0: detected conn error (1022)
Feb 28 14:11:21 pve3 iscsid: Kernel reported iSCSI connection 6:0 error (1022 - ISCSI_ERR_NOP_TIMEDOUT: A NOP has timed out) state (3)
Feb 28 14:11:21 pve3 iscsid: connection1:0 is operational after recovery (1 attempts)
Feb 28 14:11:23 pve3 pmxcfs[1709]: [quorum] crit: quorum_initialize failed: 2
Feb 28 14:11:23 pve3 pmxcfs[1709]: [confdb] crit: cmap_initialize failed: 2





 

I've seen somewhere that it occurs because multiple hosts are using the same iscsi initiator name.

So why can't i declare each controller with different iqn and do you think it's the problem i'm facing?

 

 

Thanks a lot.

 

10 Posts

March 9th, 2023 08:00

I finally resolved my problems :

1) i had a wrong Jumbo frame configuration: the array only supports 8900MTU so i put 9000 on the iscsi switches (i think i can put even more ) and 8900 MTU on the pve hosts interfaces and everything worked smoothly (The multipath conf above is right).

Thank you for your help

 

Moderator

 • 

8.7K Posts

February 28th, 2023 12:00

Darant, 

 

To start, Proxmox VE isn't tested or supported on the ME4 so that may be causing an issue, you can find the supported OS here though. 

Also, treating the connections as portals doesn't sound correct to me either as portals would be the ip of the iscsi target. I dont know if its like VMWare where you put in one and then it finds the rest for you or if you need to manually enter one for every IP of the storage. 

What I would suggest is using the Linux Best Practices guide, found here.

 

Let me know if this helps.


10 Posts

February 28th, 2023 23:00

Thank you for your answer, I know for the supported OS but i have no choice. I think you're right for the target part but i already tried  putting only one target : the one of the array, that's why i tried to enter all the Controllers but doing one or another i'm getting the same iSCSI errors.

I'll try again to put one target see what if what happens  is related to Multipath configuration.

The ME4024 device is declared as below in the multipath.conf can you tell me if this sounds right to you ?

devices {
device {
vendor "DellEMC"
product "ME4"
path_grouping_policy "group_by_prio"
path_checker "tur"
no_path_retry 24
hardware_handler "1 alua"
prio "alua"
failback immediate
rr_weight "uniform"
path_selector "service-time 0"
# no_path_retry fail
}
}

 

 

Thank you

No Events found!

Top