As an interim step, after I have data back-ups, I want to replace the failed drives. I’ve never done that, but I can follow a guide so I think I’ll be okay there.
The question is, should I replace the two failed WD Red / SMR drives in-kind? I understand now they’re not recommended for Z2/3.
Or, is it even possible to replace the failed WD40EFAX with a better drive? The idea would be to swap out the EFAX drives one at a time, beginning with the failed drives, until the whole pool of disks is a better type of SATA drive.
With SMR drives a resilver is usually the biggest killer. I’d argue that there is a chance that the drives are possibly even still alive, but this would have to be validated without a port multiplier in the mix.
If you are going to keep the existing drives, I’d first see if bringing them over directly to motherboard or HBA. I would only use CMR drives going forward, no point replacing like for like.
SMR drives usually hate a resilver & it can take an absured amount of time for one to complete. If it completes at all, and if it completes without taking out another drive along the way. It might make more sense to drop the smr drives entirely & rebuild the pool outright with cmr drives. It would very likely be faster at the very least; I think folks have reported multi-week resilver times for singular drive replacements on smr pools in the past.
Edit: it might be worth looking into burn-in guides & getting familiar with tmux and badblocks. It is generally considered good practice to make sure every block is working in a new drive before deploying it in a pool. This takes time (6 days for 1 8tb drive or so, but can be run on multiple new drives at the same time), but saves headaches in the long run. There should be some good guides on the forum. Take extra special care when running badblocks if you do decide to burn-in new hardware. I’m personally so parinoid about making a typo with badblocks that I only run it on systems that have 0 data I care about on them.
Okay, that’s a plan. I have two pools, the EFAX/WD Red with 5 drives, and a pool with 6 drives (ST3000NC002-1DY166). Is the Segate Constellation ES a good choice? Gee, I hope so! These are advertised to 24/7 enterprise use, specifically for cloud/data center and NAS use. Did I get the wrong drives twice???
I can pitch the WD-EFAX / SMR drives and copy data onto the Seagate Constellation Drive Pool?
After that I have a bunch of 6GB/S SAS drives, those are also Seagate Constellation ES-2. For that controller I have an LSI SAS card that fits into a PCIeX8 slot. 9207-8i PCIE3.0 6Gbps HBA LSI FW:P20. I’m hoping that will replace the pool of WD-EFAX drives, and provide some redundancy until I get another NAS or two online?
You know, this whole data cache that I manage is very small compared to anything enterprise. I’ve just calculated that I don’t want to lose around 3TB in data. That number will quickly expand in the near future because we all use and save more video. But for now it is less than 3TB
Google provided no clear answers. I’m 95% sure these are cmr drives, but can’t find anything to specifically quote manufacturer on this
Maybe it is worth asking their support to confirm. I’ve generally noticed that any questions in regards to enterprise grade hardware receive much better support than average, even for quick questions.
Edit:
After installing the lsi card, run sas2flash -l (or maybe -list) to confirmware firmware version & that it is flashed to ‘it’ mode. Otherwise we’ll have more work ahead of us
These are “Cloud Storage” disks explicitly for NAS. I also cannot find anything which states whether they are CMR or SMR.
Assuming that your pool is running OK but degraded, you should be able to replace drives and resilver. If you buy CMR replacements for the SMR drives, the resilver should be writing ONLY to the new drive and reading from the others which should hopefully be fine for SMR drives (though still stressful).
