I've got a setup here at work that is killing me! We've got 2 Dell R410's hooked up to a DroboElite for iSCSI support. The Drobo wasn't my first choice, but the powers that be demanded we get something certified so....
Anyway, my problem is this: basically any time the VM's start doing anything intensive, the Drobo apparently locks up. All of the datastores disconnect from the servers, the Drobo won't show up on the Dashboard, and I have to power the unit off and back on again to get it to work. I had this setup briefly with 4.0 and didn't have any issues at all with the setup, so I don't think it's a faulty unit. However, since we're going to Windows 7 on all of the boxes, we had to upgrade to 4.1 to get things to run smoothly (4.0 update 1/2 isn't an option because they are only providing 4.0 no updates or 4.1 as options for Academic users).
I've been on the phone several times with the guys at Data Robotics, and haven't had much luck yet. So far they've had me reset the unit to factory defaults to make sure the 4k cluster drives aren't causing an issue, and asked me to change the timeout settings in vSphere. The timeout settings command they gave me was: esxcfg-advcfg -s 14000 /VMFS3/HBTokenTimeout
I've tried that in 4.1 and it doesn't work. Does anyone know how to change the timeout in 4.1, or have any experience with this setup and have some tips they can give me? The only other option I've considered is taking the iSCSI off of the Broadcom hardware NIC's which use TOE, and go back to straight software initiators. I'd probably have to reload the server to do that though, considering the stupid hoops I had to jump through to get the hardware initiators to work in the first place. I'm not sure I can put everything back the way it was before.
Anyway, my problem is this: basically any time the VM's start doing anything intensive, the Drobo apparently locks up. All of the datastores disconnect from the servers, the Drobo won't show up on the Dashboard, and I have to power the unit off and back on again to get it to work. I had this setup briefly with 4.0 and didn't have any issues at all with the setup, so I don't think it's a faulty unit. However, since we're going to Windows 7 on all of the boxes, we had to upgrade to 4.1 to get things to run smoothly (4.0 update 1/2 isn't an option because they are only providing 4.0 no updates or 4.1 as options for Academic users).
I've been on the phone several times with the guys at Data Robotics, and haven't had much luck yet. So far they've had me reset the unit to factory defaults to make sure the 4k cluster drives aren't causing an issue, and asked me to change the timeout settings in vSphere. The timeout settings command they gave me was: esxcfg-advcfg -s 14000 /VMFS3/HBTokenTimeout
I've tried that in 4.1 and it doesn't work. Does anyone know how to change the timeout in 4.1, or have any experience with this setup and have some tips they can give me? The only other option I've considered is taking the iSCSI off of the Broadcom hardware NIC's which use TOE, and go back to straight software initiators. I'd probably have to reload the server to do that though, considering the stupid hoops I had to jump through to get the hardware initiators to work in the first place. I'm not sure I can put everything back the way it was before.