Drobo keeps dropping iSCSI connection

I have the DroboPro connected to a VMWare cluster using iSCSI, and it works fine for a while, but when I start copying multiple large files to it, it locks up or disconnects or something. I can still ping it, but I can’t access it on the dashboard or through iSCSI. I can’t reconnect until I reboot the device.

I have the latest firmware, and it’s formatted in 2tb partitions.

Any ideas?

Have you looked at: http://www.drobo.com/pdf/DroboElite_VMware_Best_Practices.pdf

If you are experiencing connection losses, especially during intensive operations, change the default iSCSI timeout on your ESX server to 14 seconds. This can be done using the command:

esxcfg-advcfg -s 14000 /VMFS3/HBTokenTimeout

Thank you, I will do that. Are you sure it’s “-s 14000” not “-s 14” in the command? And does that require a reboot of the ESX server?

Also, when this happens, my Windows 2008 server that’s connected via the iSCSI initiator also loses the connection. Is that related? Is there a way to increase the timeout on Windows too?


That command was given to us from VMWare, so I’m assuming yes it is 14000.

IF you are still having issues after the command change, please open a support case.

One more thing. The Drobo is connected to a dedicated switch for iSCSI traffic, which has jumbo frames enabled. I can’t change this, because the ESX hosts communicate with our Sun NAS with jumbo frames.

I can’t understand why you don’t support this feature on a nas. Is there no way to enable this on a DroboPro?
Are there any known issues on the Drobo and jumbo frames?

Jumbo Frames is not supported on the Pro. It was not designed for it. The Pro is not a NAS.

The Elite was designed to enable Jumbo Frames.

I know you’re not the designer, and it’s not your fault, but I have issue with Drobo’s decision on this. How can it not be designed as a NAS? Its prime feature is iSCSI. It’s certified for VMWare, and can be shared by multiple ESX hosts. Why would Drobo go through the trouble of VMWare certification for a simple direct attached storage device? It’s even rack-mountable, implying that it’s enterprise-grade. It’s clearly a NAS that has a few features disabled and a poor UI just so it can be called a DAS.

I’m not asking for major features to be added. For a $1500 product, I think you guys could enable jumbo frames. Many generic network cards support it, so I’m sure it can be done via a minor firmware update. Even as a DAS, jumbo frame support could greatly improve network performance.

@soy lent, your tone toward @Jennifer seems a little harsh. For what ever reasons dri decided, the Pro doesn’t have jumbo frames. The droboelite does, as does the new nas box drobofs. I don’t understand their product strategy, but they do seem to come out with new products rapidly, with each building on features from others.

Actually, the post was directed “@” DRI, not “@” Jennifer. And the FS does not support iSCSI.

I have exactly the same problem as this on OS X (Xserve running 10.6.3, 64bit). Direct iSCSI connections to two Drobo Pros (two separate ports, no switch), and if I transfer any moderate amount of data (doesn’t need to be much, a couple of GB), the iSCSI connection is dropped.

This isn’t using VMWare, just directly connected to the Xserve and using Drobo’s iSCSI initiator for OS X, and it’s driving me MAD!

Then please open a support case.

I’ve already opened two, haven’t had a particularly useful response as of yet though …

What are the case numbers?

The latest case (submitted a few days ago, with the units connected over iSCSI) is 100514-000106, I’ll get the other as well for you (that was when it was connected via Firewire). So far I’ve submitted the diagnostics log and other info, and have then been asked to submit the diagnostics log and other info (paying attention there I see!). Will see what happens this time.

So far you’ve responded far quicker and more often than anyone from the support team!

Please keep in mind we are only open between 6am and 6pm Pacific time Monday through Friday. (Your tech actually works Sunday through Thursday.)

And we do need you with your Drobo in order to troubleshoot.

Also, I have the luxury of being able to read and update the forums while on the phone and working cases from our queue. :wink:

(Also little FYI about your case, you had no attachments, that’s why we asked for the diags and system logs. We normally don’t ask if they are already attached to your case.)

Do you know how long it’s likely to take once we get pushed up to “Tier3”. The issue has happened again this morning but I’ve not had any updates since it was escalated and don’t know if we’re talking days, or weeks, for a response?

Usually 2-3 business days.

Well, nothing yet so it looks like it’s going to be next week. Are you able to see if Tier3 support have started to look at the problem yet?

No, I do not have access to their queue.