iSCSI on Mac OS X Yosemite Clients not working for past 2 weeks

[Dave Genton]

Last week and week prior I noticed my OS X clients using Global SAN iSCSI Initiators starting ejecting the iSCSI LUN about ever 20-30 seconds. I had not time so just ejected and turned off persistence. Returned back seeing another update, then two more hoping those would fix it but no go. For past 2 weeks I have had no access longer than or as long as 30 seconds on my OS X clients. VMware server using iSCSI LUNs are fine and no problems.

Any known changes to iSCSI affecting these Global SAN OS X clients ?? Anyone else having these issues ?? I've been using these LUN's and clients about 2 years on this freenas box without issues until an update 2 weeks ago which has now gone on too long to ignore any further. All I am noticing on the FreeNAS side is constant logging in window of a child process terminating, hadn't looked into it hoping any changes would be fixed. Made no changes on my side besides the FreeNAS auto-updates being pushed out, since went in and gone thru iSCSI setup, changed portals and all that, no go. Vmware rock solid, OS X / Global SAN no go at all.

How shall I proceed to find issue ??

thanks in advance,
dave

 

[Dave Genton]

Post down below by JANO has same log errors mine has..

 

[dlavigne]

Which version of FreeNAS? Anything related in /var/log/messages on the FreeNAS side?

 

[L]

I have all kinds of trouble with the globalsan iscsi initator and yosemite. I downloaded a free iscsi intiator from small-tree and it works great. very stable.

 

[Dave Genton]

Which version of FreeNAS? Anything related in /var/log/messages on the FreeNAS side?

Using STABLE 9.3 train, and as I stated fine up until 2 weeks ago, three auto-updates. Nothing besides same messages posted by JANO in his post this week I pointed out. I have never had issues with Global SAN in the past and as I am sure you know these are licensed initiators I am using on OS X 10.10.2. Today I have one client running with target connected while monitoring the log to look for anything new. I've been having to leave it disabled as the LUN just pops up and down so much now.

 

[Dave Genton]

Interesting, logs are still clean, not even getting the messages JANO is getting any longer. LUN would just eject and reappear over and over and over. I disabled error detection and it stopped. I've always used "Header and Data (recommended)" since using this combo a couple of years ago. Today I set it to "off" and it mounted the LUN and hasn't dropped since. Oddly I can up the error detection to header, or header and data now and its staying up and transferring data. The initial connection for some reason now must have error detection disabled. If not it will go up and down every ~30 seconds or so until its disabled. Will run like this awhile and file bug report shall this stay consistent but I'd like to see something in the freenas logs ..... nada, zilch right now.

d-

 

[Dave Genton]

Its back to doing it again and these are the only things I see logging on Freenas.

Feb 15 16:02:20 FreeNAS ctld[77185]: 192.168.1.42 (naa.bd79fec83b074c24979584ddeb9e2861): read: connection lost
Feb 15 16:02:20 FreeNAS ctld[2307]: child process 77185 terminated with exit status 1

Just cycles non-stop, mac side only see lun eject, then appear immediately. This same thing others are seeing since recent 9.3 stable update ?

d-

 

[dlavigne]

Please create a bug report at bugs.freenas.org and post the issue number here.

 

[mav@]

Dave, if you can so quickly reproduce the problem, could you collect packet dump of such problematic session with `tcpdump -pi <interface> -s 0 -w <file_name> port 3260`. Unfortunately I have no Mac to reproduce the problem myself.

 

[Dave Genton]

Please create a bug report at bugs.freenas.org and post the issue number here.

Issue #8058 created.

 

[mav@]

The bug report is good, but what's about packet dump?