Starwind Free Edition (latest build) causes server restart

[jmchristy]

I'm posting to see if anyone has heard of or experienced this behavior.

My setup...
2 PowerEdge 2850's with Starwind latest build HA free - 4 SYNC channels setup each 1GBe, jumbo frames enabled
2 PowerEdge 2950's running ESXi 5.0 with all the patches/updates
2 iSCSI data ports out of each server/SAN, 1Gbe, jumbo frames enabled


I'm here working over the weekend, attempting to roll out this HA Free edition at a remote site and utilizing the 128GB HA - and whenever I attempt to add the HA target to VMware 5.0.914586, vCenter build 5.0.0.913577 - one of the two Starwind SAN's reboots. I have replicated this now twice, the second time I allowed the SAN to reboot, come back up, SYNC'd, and I attempted to move a virtual machine to the target and one of the SAN's rebooted again. It's never the same SAN, first time it was the secondary SAN and the second time it was the primary.

I replicated my setup that I have already rolled out at my corporate site, just with the latest Starwind build. I've been trying to figure it out most of the day, tried HA targets at 128GB of size, 120GB of size, and 100GB of size - i thought maybe Starwind was recognizing more than 128GB in use and was rebooting.

I have since went down to Starwind version 5.8 - which I was running previously on this exact same hardware and it's working ok. Just no HA.

I'm not currently running Starwind HA latest build anymore - but if I can still pull old logs from some directory I will. Time is not my friend unfortunately, I only have until Monday morning - I don't think my fellow associates would appreciate coming in Monday morning to a crashed network

[anton (staff)]

1) StarWind is user-land application so it cannot make machine BSOD by itself. If it happens it means something within either storage or network stacks (ones we
do interact with) is broken and we do somehow exploit this "opportunity".

2) We'd be happy to help but we do indeed need both logs and at least mini dumps from a failed machine.

[jmchristy]

Yes, you are correct.

I came across this post
http://www.starwindsoftware.com/forums/ ... t2794.html

I am running a very similar setup and the issue is with the Intel Pro 1000/MT network adapters in Windows 2008 R2. They do not support Jumbo frames - even though there is a setting to enable them it doesn't work. I couldn't ping anything using ping -f -l 9000, or 4088 even. I was able to get ping replies @ 4000, but I still had to disable Jumbo frames on the SYNC channel's, enable RX/TX flow control, and set the NICs to 1000MB instead of auto.

I have since reloaded Starwind free HA, and I haven't experienced the behavior that I posted yesterday. I do still have Jumbo frames enabled for the iSCSI traffic because I can specify 4000 in VMware.

[anton (staff)]

I'd recommend you to swap these Intel's for something known working with Jumbo frames flawlessly. Upgrade should come cheap these days.

Yes, you are correct.

I came across this post
http://www.starwindsoftware.com/forums/ ... t2794.html

I am running a very similar setup and the issue is with the Intel Pro 1000/MT network adapters in Windows 2008 R2. They do not support Jumbo frames - even though there is a setting to enable them it doesn't work. I couldn't ping anything using ping -f -l 9000, or 4088 even. I was able to get ping replies @ 4000, but I still had to disable Jumbo frames on the SYNC channel's, enable RX/TX flow control, and set the NICs to 1000MB instead of auto.

I have since reloaded Starwind free HA, and I haven't experienced the behavior that I posted yesterday. I do still have Jumbo frames enabled for the iSCSI traffic because I can specify 4000 in VMware.