One drive dead and could use a recomnedation which way to go.

[Dirk]

After 5-6 years One of the 6 2TB drives has died in the old server that was recently re-tasked. When I repurposed it to 5 drives in Z2 and a hot spare. If I have to buy another drive it's silly to get the same size BUT And here is where some advice would be appreciated. Given that the MOBO is maxed at 16GB (given the recommendation of two GB memory per TB) what are my options?
Go with replacing the dead one with a 3TB and do the rest as they die or go with RAIDZ 4 drive and hot spare and start over. I just don't see it prudent to throw another MOBO and drives at the problem so I can go with larger drives on an already dated system.

I just did a reboot to see check for lose cables and at restart the swap drive was brought in. With six drives in there how can
I tell which one is failed? It no longer shows up under disks.

Thanks in advance. Specs on servers are below.

 

[ChrisRJ]

The recommendation is 1 GB RAM per TB storage, IIRC

 

[sretalla]

RAM:TB storage is just a rule of thumb, not a real "requirement".

You can run a pool on much less than 1GB per TB. (just don't expect impressive performance from it)

With six drives in there how can
I tell which one is failed? It no longer shows up under disks.

Look at the serial numbers in the disks list or in dmesg | grep Serial and find the missing one in the actual disks you have in the box by process of elimination. (keep a list somewhere for future reference to link the positions of the disks with their serials to avoid this messing around next time)

 

[Dirk]

The recommendation is 1 GB RAM per TB storage, IIRC

Thanks Chris. Looking just now I found more info on GB memory VRS number of drives than GB memory VRS TB of space.

sretalla: IT's a media server. No VMs and one user (maybe an occasional 2). Nothing special going on other then three jails with media related software. I think any load would be more CPU (for transcoding) then memory related.

If you agree and nobody else inputs I'll start replacing the 2TB with 4TB 5 in a z2 and a hot swap which is what it's at now.

Last question, when the failed drive is replaced will the pool make it the hot swap (which was brought into the pool replacing the failed one), or pull it into the pool in place of the removed one?

 

[sretalla]

Last question, when the failed drive is replaced will the pool make it the hot swap (which was brought into the pool replacing the failed one), or pull it into the pool in place of the removed one?

General answer... the system won't make the decision for you, so you need to do something to either: replace the spare in the "replacing" status and return the spare to the pool or leave the spare in its new place and replace the spare (if you still want one).

Where you can, you should follow the GUI process, but I think this will require some CLI depending on how you want to go in the end.

https://www.ixsystems.com/documentation/freenas/11.2-U5/storage.html#replacing-a-failed-disk

For the CLI part, you need to decide if you will permanently use a spare in the pool, replace the disk in the pool and return the spare.

This article (although it's strictly not Open ZFS like we use on FreeNAS) covers the process.
https://docs.oracle.com/cd/E19253-01/819-5461/gcvcw/index.html

 

[Dirk]

Thanks, I don't see any sense in replacing the drive with a larger one and have it sit there as a hot swap. Dropping one of the 2TB back to a hot swap made more sense to me.

Thanks for the links, I'll read up.

In regards to serail numbers and finding the failed drive, I use Node 304 cases which use a two pack white cage for the drives. I began using a sharpie on the cage to note the serial#.

 

[Dan Tudora]

hello
Can use smart_report.sh from https://github.com/Spearfoot/FreeNAS-scripts to find more info about your HDD including serial number
If you make a cron job weekly you have in email info about HDD weekly
Of course long and short SMART test MUST to be configured to call you your Free/TrueNAS (email) about health of your HDD
Recommended (from forum read) is to do short SMART daily and long weekly at night (when the server is in not in load)
Avoid to make a long or short SMART when is scrub programed

cheers