Seagate Nytro - Predictive Disk Failures.

Posted by sigserv@reddit | sysadmin | View on Reddit | 12 comments

We are an MSP. We have quite a few clients now where the RAIDed drives on their servers, that have the Seagate Nytro model of disks, come back with predictive disk failures upon random server restarts. All or a majority of the disks are either 980 gig or 1.92 TB Seagate Nytros. The servers are either HP or Lenovo servers.

So now we have a few servers populated with disks flagged with predictive disk failure. They are mostly all out of their 5 year warranty window. Normally we would simply order replacements. However the enterprise disk market out there is stupid expensive.

Has anyone experienced this and does anyone have a workaround for this?

[-]

udsd007@reddit

In addition to all the other advice, buy discs in small batches, made in different plants, if you can. It is pure Hell when all the bathtub curves coincide.

[-]

rhubear@reddit

Reddit is littered with people with Seagate problems.... Most posts are in the Seagate Sub.

My advice.... stay away from Seagate.

Years ago, my then storage dealer had a customer with a Seagate HD problem. The HD was returned to Seagate. Seagate refused the return. The hard drive subsequently completely died at the customer.

The storage dealer did two things :

He blew up at Seagate.
He blacklisted Seagate.

At the time, he migrated all his customers (inc me) to Hitachi HD.

Years later Hitachi was discontinued by Western Digital. So approximately 15 years ago, I stopped using Seagate's HD. I used Hitachi and then Western Digital Pro drives (Red Pro/Plus or higher model).

I've forgotten what an RMA is.

[-]

notospez@reddit

Just do what we used to do in the dark ages before SMART. You wait for the drive to fail, and regularly check your backups. That's what RAID is for...

[-]

brokenpipe@reddit

Surely you have spare disks on hand as a responsible MSP.

[-]

AniBMagal@reddit

Monitor SMART attributes manually.

[-]

buck-futter@reddit

Be aware some controllers outright lie and report attributes that are a creative interpretation themed loosely on the drives own values. I've had a couple of Dell servers specifically that claimed a drive was written to death and about to fail, we replaced the drive and rebuilt the array but on checking the disk in a different system there were no reallocated sectors and the drive itself was not in SMART failure predicted.

[-]

Calleb_III@reddit

If you are an MSP, but who owns the kit, you or the customers. If it’s the former, why haven’t you priced in replacements after the manufacturer warranty is over? The storage price increases were widely expected 12-18 months ago. If you didn’t stock up that’s on you (as a company).

You can replace only Tier1 systems preemptively and move the predictive failure drives to non-prod etc. up the RAID level a notch and prey for the AI bubble to pop, before your drives pop

[-]

Beach_Bum_273@reddit

Which MSP, so we can add you to the list of MSPs not to use

[-]

Helpjuice@reddit

There is no workaround, you are in enterprise stop using disks outside of their warranty and get stock in place to replace them before they fail. The predictive failures are accurate and are what you will be running into if you do not operate things like an enterprise business rotating out the old with the new on regular intervals which is max 3-5 years.

The pricing should be in the budget, some other non-critical business wants that others want to do needs to be cut or pushed back in order for operations to succeed to be maintainable or data loss will occur for your clients.

This is the cost of business, adjust prices accordingly or suffer from disk failures that could have been prevented.

[-]