Disconnecting and Reconnecting VMware event

Hello out there!

I am reaching out to the community in hopes to get some help with an issue I am having with my new Drobo b1200i. I’ve reached out to Drobo directly via their support staff and replies are pretty delayed. I am hoping some one might have had this same experience and resolved it. I found some vmware forums of people who had this same issue - and the end result of the ten page forum was that most people returned their Drobo.

So - my Drobo in all of its glory went into production just over ninety days ago as a datastore appliance for my vmware environment. One day I was doing some general perusing of the event logs on one of my ESXi hosts and I noticed the Drobo was reading the below (shortened the error up a bit)

“Lost access to volume… due to connectivity issues… Recovery attempt in progress…”

A few times a minute just disconnecting and reconnecting repeatedly as far back as the log could see. So I storage vmotioned everything I had in production off of it - plugged in two more network cables, assigned IP’s (was only using one iSCSI port at the time), rebooted it, enabled multi-path round robin in vmware and migrated some vms back to it and confirmed the paths had active I/Os on each port. It seemed to lessen the instances but still many disconnect/reconnects in ESXi event log.

Oddly - the issue never actually triggers a warning or alerts (just events) and the VM’s themselves never show as disconnected. It appears that the OS’s on the VM’s are able to tolerate the latency.

Drobo is configured with RAID 5 and using 3 SSDs application aware feature. Its has 3 iSCSI connections to a switch (the same switch is used for 3 other storage appliances for this ESXi cluster. Different manufactures on those - none having connectivity issues and the switch is grooving along just fine).

At present I have migrated all the VM’s off the drobo accept for 3 small ones. The connection issue is still there but it is less frequent - but I expected the drobo to be able to handle a much heavier load.

Anyone experience this issue? I have been a fan of Drobo in their products simplicity for many years - not had an issue with any product ever. When I had the chance to purchase a new storage appliance and I saw Drobo had certified the 1200i for ESXi 5.5 I was stoked - so pretty down in the dumps on this at present.

Thanks in advance random person of kindness ; )

hi jwinders,
im afraid i do not have much experience with vms and that model, as most of mine are Das models via usb, though i was just wondering something…

if all of the vms, do not actually have any connection problems storing data onto the drobo, or reading from it etc… what if the log is somehow related to the status polling that dashboard (or its service) does with the drobo?

are there any other client/host-specific logs that you could access to investigate further, such as default windows Event collector/Event logs, (or ones that you could possibly enable for some testing)?

its always possible that something is causing those errors (which shouldnt be there) but am also just wondering if perhaps the virtual nature of it is somehow causing a false positive in some way? (sorry if i misunderstood anything though)

Hey Paul,

Thanks for your post! I appreciate you taking the time to read through my issue.

Is there a way I can turn off this polling from the Dashboard to test? Although when I noticed the issue I did not have the management network connected - so the dashboard (which I imagine is the source of the polling?) could not have reached the Drobo.

I did not consider any client side logging because no matter what flavor of virtual machine I put on there the issue happens. I would imagine any logs, since its transparent to the VM’s that they are virtualized, on the client side will show I/O and disk latency related messages if anything.

The issue seems to persist regardless of what VM I put on it - and if I migrate the VM to another Data Storage appliance - say my EMC StoreCentre or my Dell Powervaults the issue disappears.

Thanks again for your post man - if you let me know where that polling feature is I will turn it off to give that a shot.