Bad sectors during initial zfs send / recv -- replace drive now?

[Z300M]

I am setting up a new RaidZ2 pool consisting of seven brand-new ST8000DM004 drives. All (plus a spare) passed initial S.M.A.R.T. (short, long, and conveyance) and badblocks testing, but now, while populating the new pool by means of zfs send / recv from an over-full smaller pool, I'm getting a report of 8 unreadable / uncorrectable sectors on one of the new drives.

Quite apart from the issue of whether I return the drive (and buy a new one at a now-increased price, if they will not simply replace it without paying more) or RMA it with Seagate (and receive a "refurbished" -- i.e., more thoroughly tested -- replacement), would it be OK to replace the drive by the spare now, or should I wait until the new pool has finished populating?

 

[Z300M]

Later: Now it seems that the operation has terminated. I've seen no increase in the data size of the new pool, and the drive activity LEDs are not indicating any activity. AND I did not use the parameters that would have allowed resumption of the send /recv operation. BUT an answer to my original question might be useful on some future occasion.

 

[rds13]

I have also recently build a fresh freenas server with seven brand-new ST8000DM004 drives. I also passed all the same disk tests which took more than a week to complete on each disk. No error were found as far as I understand the report.
I also made a copy of an existing ZFS filesystem from another server via zfs send and 2 disks started to report 8 uncorrectable sectors. One now report 16 uncorrectable sectors.
I am really disappointed as I never have taken so much care to build a server before and the result is below my expectations to say the least.
I'm looking for advice. Should I buy seven other disks but not from Seagate ?

 

[Johnnie Black]

seven brand-new ST8000DM004 drives

If nothig else you should avoid this model, as these drives are SMR and not recommended.

 

[Z300M]

If nothig else you should avoid this model, as these drives are SMR and not recommended.

I get the impression that many high-capacity "consumer-grade" drives are SMR these days. If the caching is adequate, does it matter much?

 

[Johnnie Black]

If the caching is adequate, does it matter much?

Yes, there are several threads with users with the new WD SMR disks having terrible performance, especially during resilvers, all new WD except 2 and 6TB are still CMR, all 3.5" Toshibas are CMR, most Seagates except Archive and some Barracuda are CMR, in fact and AFAIK all disks larger than 8TB are CMR, except the 14TB SMR HGST but that one is host managed and not availbale in retail.