Drobo

Question regarding DroboPro memory of failed drives

Does DroboPro have some type of internal memory that records the SN of drives that are reported failed?

How does the DroboPro determine that these hard drives actually failed? Is it based on one incident of failure are repeated testing of that drive once its reinserted?

A few of the drives DroboPro reported has failed works perfectly inside of my computer (tested with large amounts of data/SMART is verified/no issues). I have my workstation connected to a UPS and had a report of a outage today, DroboPro then reported one drive was failed. I backed up all data on the DroboPro and did a reset… now everything is fine. It’s not reporting a bad drive.

I just want to know that these drives are indeed really Kaput before I go through the hassle of sending them back to WD. One drive that was reported failed has been working in the DroboPro for months now with no issues.

If it works the same as Drobo (which it likely does), then yes.

There are other threads discussing “premature” bad-marking, but my opinion is that Drobo is more sensitive to drive issues and provides an early indicator. Some may consider it paranoid, but if you weren’t at least a tiny bit paranoid about your data, you wouldn’t have a Drobo to begin with, right?

Unfortunately the manufacturer-supplied drive diagnostics pretty much only fail the drive when it has exhausted its replacement blocks, or it has a completely unrecoverable fault.

Drobo, on the other hand, does its own error-monitoring and reallocation and it seems that once a drive has enough errors, even possibly recoverable ones, then it marks it BAD.

There’s also the possibility that TLER is conflicting with Drobo’s error handling and causing false-positive timeout errors due to both the Drobo and the drive trying to reallocate blocks.

As for SMART, while it’s an indicator, it is not a surefire way to detect a drive going bad. I had a drive that completely failed and never reported a SMART status of BAD. Looking back, it was obvious the drive was going bad - it was “stalling” the system and failed to spin up from cold, but again, SMART never reported a problem.

On another machine a similar thing happened and SMART reported BAD only on the last boot-up before the drive went into a nasty clicking death. Luckily I was able to save the data, so I thank SMART for that. But its warning was surely not early for me.

So I liken this to having a dish that shows a crack, but has not shattered.
Drobo says “No way, I’m not using that - it might break at any minute” while other things say “Aww heck, I’ll keep using it until it actually shatters.”

Brandon

Thanks Brandon for your input. I would still like to know how Drobo calculates if a drive is bad. How comes it marks a drive has bad but when you reset it and install the same marked bad drive the system works perfectly for months? I don’t think Drobo should have a memory of bad drives. Seems to me that it is reporting perfectly fine drives as bad. This has already happened a few times already and all the drives are working perfectly on a Drobo that is 75% full.

Give us some option to reuse the same drive in the system by allowing us to reset the internal list of bad drives. is this already possible without having to reset the Drobo and losing all your data?

Part of the problem is that we (as customers) don’t know exactly what criteria Drobo uses before it decides a drive is bad.

If the drive’s working fine in another Drobo, it sounds like you’re getting a false-positive.

If you send DRI your log file, they can analyze it and tell you why it’s marking the drive bad - maybe even analyze against the log from the unit that the same drive is working OK in to see what the trigger is.

IIRC, a pin-reset of Drobo without the drive pack inserted (power down Drobo, eject all drives, power up Drobo, perform reset, power down Drobo, reinsert all drives, power up Drobo) will reset the bad drive list, but I would wait for another member to confirm this before trying it.

Yes… a big part of the problem is that we are left in the dark has to what events will make a drive listed as bad.

The drives work fine in the SAME DroboPro after a reset, dumping of files to 75% full and months later of operation.