Jails on SSD?

[Daisuke]

Hi,

I have a C2100 with 2x2.26Ghz procs, 96GB RAMD and 12x2TB disks. I'm installing 2 SSDs, one for the OS (64GB) and one for jails (128GB). I believe there is an an option to have redundancy even if you only get the one SSD. I read that you can set copies=2/3 for any datasets you want, to be less likely to get corrupted. Do you have a guide or a procedure that would help me achieve this?

Thanks.

 

[sretalla]

zfs set copies=2 <pool/dataset>

Not sure you will gain a lot by doing that, but it will provide the possibility that a scrub can identify a problem with one of the copies and repair it with the other.

Much better would be to replicate it to a larger, spinning disk and have snapshots to roll back to.

 

[Daisuke]

This is a fresh setup, nothing is configured yet. So technically, I have 12x2TB hds and 1x128GB ssd to configure. I read the documentation and there are no suggestions how to setup jails taking advantage of an ssd.

If you guys have any suggestions, please share, I would like to set everything properly the right way. This NAS will be used only to store my personal photos and YouTube videos.

 

[hervon]

Plenty of examples in the forums. I built a pool with 2 x500gb ssd miroir. Works great!

 

[sretalla]

If you guys have any suggestions, please share

Create a pool on the SSD you want to use.
When setting up the jails repository (on first jail creation or from the GUI before that), select that SSD pool.

I recommend making sure when you set up your jails that you segregate out the data and config components to a dataset like ssdpool/apps/app1data, ssdpool/apps/app1config, etc. Then mount those into the appropriate locations in the jails.

That way you can set different replication/snapshot intervals for the jails themselves and the data/config. You could then replicate those to the spinning drives pool to have the protection against SSD corruption and only 1 copy.

 

[BBarker]

I actually mirrored two 480gb SSDs for my jails pool. Noticed that SABNZBD would cache downloads on them until they could be extracted to their final location in my media dataset located on spinning drives. My original setup using 120gb drives didn't have enough space on some larger downloads and would stop mid process.