Project

General

Profile

Bug #6296

iSCSI target going offline

Added by David Bond about 4 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
-
Start date:
2015-10-07
Due date:
% Done:

0%

Estimated time:
Difficulty:
Medium
Tags:
needs-triage

Description

2 x Dell r720xd running Omnios (r151010) not tested on r151014
2 x Dell r710 running ESXi 5.5, with 2 windows 2012 R2 VMs

Storage network is 40Gb inifiniband, 2 swicthes, linked, different partitions for each iSCSi path and 1 for NFS, each system has 2 ports, 1 to each switch.
The 2 Omnios systems are running with shared storage, with RFS-1, pool running on each, sharing out NFS and iSCSI targets.
2 x windows 2012 R2 VM created and running on an NFS share with in VM iSCSI for use with microsoft clustering.

If you setup the Windows VM a with iSCSI and an iqn, enable MPIO, and create 2 sessions per target (1 target from each Omnios system) and log it into the iSCSI target on each on the Omnios systems, the client will make 2 sessions per target and see the luns as expected.
If you now set up the second Windows VM, give it the same iqn (by misstake), enable MPIO etc the same as the previous (just different host name and IP addresses) and log it into the targets, within a minute or 2, the number of sessions on Omnios will have increased to 1000's and the target will go offline. NFS will still be responding, but the target that was logged into is no longer working. You will also be unable to export the pool.

If you disconnect the second system within seconds of connecting it it will maybe increase to a few hundred sessions, the session count will not decrease after the second VM is disconnected and within 5-10 minutes the target will again go offline and you will be unable to export the pool.

Also available in: Atom PDF