carl0s
Dabbler
- Joined
- Sep 16, 2012
- Messages
- 32
Hi. I am planning to flatten my Centos 5.7 Bacula box (poweredge 2900), and turn it into a FreeNAS box. Bacula's file-volume management/labelling and general lack of ease of use is a pain for me, plus the O/S is old and unupgradable, and stuck with ext3 on the 1.5tb volume.
I think I will get on a lot better using FreeNAS with rSync over the WAN, CIFS over the LAN, and scheduled ZFS snapshots for history (in place of full/incremental/differential backups over the WAN).
It's a shame there isn't a "full install" option available for FreeNAS any more. It'd be nice to take the opportunity to play with FreeBSD a lot more. Anyway that's a separate topic.
The system consists of 3x72gb 15,000rpm SAS drives, and 3x1.5gb desktop class SATA drives, on SAS interposers, all through a Perc 5/i.
I am planning to de-RAID everything - run as individual disks, but still through the Perc5.
If all the drives were the same capacity, I would run a 6 disk raidz2.
Instead, I think I am going to build a 3 disk raidz1, and then extend it with another 3 disk raidz1, so that I have a single larger volume that allows some degree of extra redundancy, i.e. it would survive two failed disks as long as they are from the separate vdevs.
Do you agree this is the best way for me to utilise what I have, for what is essentially going to run as a backup destination. Performance is not a priority. Reliability and making use of the hardware is more important.
Here's the current disk layout on the Centos 5 box:
Now that I think about it a bit more, I'm not sure if it would just be better and simpler, to just have two separate zpools, each as a 3 disk raidz1, and use one volume for one dataset, and another for some others. Then if I was struck with a SAS drive failure, and decided I didn't want to spend the money on another 15k SAS disk, I could just move the data over to the SATA volume and retire the SAS volume, or recreate it as a mirror. Hmmm.
Any thoughts?
(p.s. I intend to install FreeNAS itself onto a USB stick or something, so all those disks are available for storage)
cheers :)
I think I will get on a lot better using FreeNAS with rSync over the WAN, CIFS over the LAN, and scheduled ZFS snapshots for history (in place of full/incremental/differential backups over the WAN).
It's a shame there isn't a "full install" option available for FreeNAS any more. It'd be nice to take the opportunity to play with FreeBSD a lot more. Anyway that's a separate topic.
The system consists of 3x72gb 15,000rpm SAS drives, and 3x1.5gb desktop class SATA drives, on SAS interposers, all through a Perc 5/i.
I am planning to de-RAID everything - run as individual disks, but still through the Perc5.
If all the drives were the same capacity, I would run a 6 disk raidz2.
Instead, I think I am going to build a 3 disk raidz1, and then extend it with another 3 disk raidz1, so that I have a single larger volume that allows some degree of extra redundancy, i.e. it would survive two failed disks as long as they are from the separate vdevs.
Do you agree this is the best way for me to utilise what I have, for what is essentially going to run as a backup destination. Performance is not a priority. Reliability and making use of the hardware is more important.
Here's the current disk layout on the Centos 5 box:

Now that I think about it a bit more, I'm not sure if it would just be better and simpler, to just have two separate zpools, each as a 3 disk raidz1, and use one volume for one dataset, and another for some others. Then if I was struck with a SAS drive failure, and decided I didn't want to spend the money on another 15k SAS disk, I could just move the data over to the SATA volume and retire the SAS volume, or recreate it as a mirror. Hmmm.
Any thoughts?
(p.s. I intend to install FreeNAS itself onto a USB stick or something, so all those disks are available for storage)
cheers :)