Page 1 of 1
ESX Server hangs when restoring iscsi targets
Posted: Sat Apr 26, 2008 3:35 pm
by WilliFixit
I am using Starwind Professional and accessing iSCSI on a Win2k3 server. I am experiencing esx server (ver 3.5.0, 82663) hanging when starting up. It stops the boot process when "Restoring s/w iscsi volumes" --> it just sits there. Other esx boxes still communicate fine with Starwind on the same subnet to the same server, even when the one cannot gets past that prompt.
(BTW, it is not just this server that has the problem, others too have difficulty during the boot up process.)
The servers are all loaded with standard configurations (identical except for IP and hosthame)
Killing the power to the server (after it has hung) and restarting seems to work sometimes.
Sometimes the machine WILL finally continue and boot up after 5 minutes at the same prompt.
Starwind has 15 targets, all image file devices.
Am I setting something up incorrectly, or is this some known issue?
Thanks!
Will
Posted: Sat Apr 26, 2008 5:04 pm
by anton (staff)
Send us StarWind logs to
support@rocketdivision.com so we could understand what's wrong in iSCSI communication process between target and initiator.
Posted: Sun Apr 27, 2008 8:16 pm
by microfoundry
I am having the same issue. I have tried various versions of ESX 3.5 & 3.5i along with various ESX patch revisions and it still is an issue.
If I bootup without the iSCSI network connected, it will process on. Plug in the iSCSI network and the server will hang.
Will, are you setting up your iSCSI as shared to utilize VMotion? I my case I am, and currently only at 2 ESX servers. But it seems to me the first server will generally startup no issue. It's only the second that hangs.
Posted: Sun Apr 27, 2008 8:37 pm
by WilliFixit
Hi,
I am using it for VMotion as well. I have also noticed that the first server in seems to work ok, but the following servers have a problem. I am working with three ESX boxes right now. I have sent the log files in to be reviewed,
I will let you know here what I find. I am concerned because I will be connecting 13 servers with the same configuration to the same iscsi box....hope that they have a fix...
Will
Posted: Tue Apr 29, 2008 7:20 pm
by anton (staff)
Of course we'll fix this as we don't have another choice...
P.S. Everything was working till ESX 3.0+ was released. So we'll find out what's broken now.
Posted: Tue Apr 29, 2008 7:26 pm
by microfoundry
Anton - is there anything I can do to help. These are non-production servers and I'll be happy to supply logs or any other info. If there is verbose logging you would like, let me know.
Thank you,
Terry
Posted: Tue Apr 29, 2008 8:25 pm
by anton (staff)
Please drop us a message to
support@rocketdivision.com referencing you're you (for some reason even spam-protected e-mail bounces back when writing to you...). Thanks!
microfoundry wrote:Anton - is there anything I can do to help. These are non-production servers and I'll be happy to supply logs or any other info. If there is verbose logging you would like, let me know.
Thank you,
Terry
Posted: Sat May 24, 2008 7:01 pm
by Nitroburn
I was having the same issue (ESX 3.5 with iSCSI Starwin on W2K3 host). I was able to resolve the issue by using a dedicated network switch for use with the NICs assigned to the iSCSI. So in other words, anything iSCSI related was on dedicated hardware (NICs, switch, IPS, etc..). I performed the reg mods and utilized jumbo frames and that cured my issue.
During ESXC boot-up, it will still "pause" at that area, however with 4 iSCSI targets, total "pause" time is about 90 seconds max.
Hope this helps.
Mike
Posted: Sat May 24, 2008 7:46 pm
by anton (staff)
Mike, thanks a lot for letting us know! Actually it's a good *workaround* but looks like we've been able to find a *solution*

Please give a try to the latest 3.5.4 RC it should work fine.
Nitroburn wrote:I was having the same issue (ESX 3.5 with iSCSI Starwin on W2K3 host). I was able to resolve the issue by using a dedicated network switch for use with the NICs assigned to the iSCSI. So in other words, anything iSCSI related was on dedicated hardware (NICs, switch, IPS, etc..). I performed the reg mods and utilized jumbo frames and that cured my issue.
During ESXC boot-up, it will still "pause" at that area, however with 4 iSCSI targets, total "pause" time is about 90 seconds max.
Hope this helps.
Mike
Posted: Mon May 26, 2008 10:32 pm
by WilliFixit
Hi Anton,
You had me try a beta version of Starwind. The file names were:
StarWindService_20080423.zip
and
StarWind_354_20080417_x86.exe
Do I have the latest? I am rolling out a server this week and want to make sure I am using the correct version.
BTW, it did solve the issue with crashing, but it still takes an ESX Server machine 4 minutes of wait time at the "Restoring S/W iscsi volumes" prompt during boot up before it finds 15 iSCSI targets on my Starwind box. And, everything has its own NIC/switch, but I am not using jumbo frames at this time.
Thanks,
Will
Posted: Wed May 28, 2008 10:49 am
by anton (staff)
Nope. Please update to the 3.5.4 release.
It's still abnormal if it takes so long
WilliFixit wrote:Hi Anton,
You had me try a beta version of Starwind. The file names were:
StarWindService_20080423.zip
and
StarWind_354_20080417_x86.exe
Do I have the latest? I am rolling out a server this week and want to make sure I am using the correct version.
BTW, it did solve the issue with crashing, but it still takes an ESX Server machine 4 minutes of wait time at the "Restoring S/W iscsi volumes" prompt during boot up before it finds 15 iSCSI targets on my Starwind box. And, everything has its own NIC/switch, but I am not using jumbo frames at this time.
Thanks,
Will