Replacing lost drive - do you need to do anything about swap?

[victorhooi]

Hi,

I have FreeNAS 9.3 Beta (will be upgraded to 9.3 Release) installed on a HP MicroServer Gen8.

The HP MicroServer has 4 SATA drive bays, and I have 3TB drives in each. I am running in RaidZ-1.

Recently, one of the drives failed - a Seagate, believe it or not. Somebody else in my apartment noticed a beeping noise, and it turns out it was the server. I'm assuming possibly the RAID controller in the server noticed the disappearance of the drive, and this was it alerting on it.

I also noticed that the web interface for FreeNAS was inaccessible - however, I was still able to access files via AFP. However, when I looked at the console output on the machine, I saw messages like:

Jan 1 00:00:00 freenas swap_pager: I/O error - pagein failed; blkno 532031,size 8192, error 6
Jan 1 00:00:00 freenas vm_fault: pager read error, pid 1 (init)​

Anyhow, I took out the drive, plugged it into a SATA-to-USB 3.0 adapter, the drive makes a few clicking noises on power-up, but otherwise doesn't seem to be detected. So it seems to have just basically got kaput, instantly...lol. Zpool status marks the volume as "DEGRADED", and says one or more devices could not be opened.

I've bought a new Hitachi Deskstar 4TB drive, and put it in place of the lost Seagate disk. I've currently doing a re-silvering process, which is proceeding along nicely.

My understanding is that FreeNAS by default creates a small swap partition at the beginning of each disk. In the WebGUI, if I go to System, Advanced, the setting "Swap size on each drive in GiB, affects new disks only. Setting this to 0 disables swap creation completely (STRONGLY DISCOURAGED)." is set to 2Gb, which I assume is the default.

My first question is, do I need to do anything special about the swap volume? Is this swap partition a separate one for each of the four drives? Or is it somehow striped across them?

Secondly, is there some kind of alerting I can setup in FreeNAS, to let me know if a volume is degraded, a SMART tell fails, or if a scrub detects any errors? I know I can configure SMTP settings under System, Email, but I can't seem to find anything to configure alerting emails.

Thirdly, I have replaced one of the 3TB drives with a 4TB drive. I assume this will not change the total capacity of my RaidZ pool yet. However, if over time I slowly replace each drive, my understanding is that after the last drive is replaced, I can then grow the capacity of my pool by using "zpool online -e" (as per http://www.itsacon.net/computers/unix/growing-a-zfs-pool-update/). Is this a correct assumption?

Finally, I have read elsewhere (e.g. http://www.reddit.com/r/freenas/comments/1ubebu/stop_using_raidz_seriously_just_stop_it/) that using RaidZ-1 is a bad idea. Apparently a former forum member, protobsd also wrote a guide on it, but I can't seem to find a valid link to that guide. With a 4-bay system, what is the current thinking around running RaidZ-1? Am I asking for trouble, or will I be OK with frequent scrubs (I believe it's set to 35 days by default, and as per above, I'm hoping to setup alert emails), and replacing drives if they go bad?

Regards,
Victor

 

[SweetAndLow]

1. You don't have to do anything special for swap
2. FreeNAS has built it email alerting, just use the email settings.
3. yes you are correct.
4. raidz1 is a bad idea with the size of current HDD's. When replacing drives you are very likely to have a read error durring rebuild causing data loss since you don't have protection during a rebuild when using raidz1.

Also you need to follow the rules and tell us more about your hardware. You mention a raid controller and if you are really using a raid controller you need to stop doing that right away. You will most likely have dataloss if you keep using it.

 

[victorhooi]

Hi,

Thanks for providing that information. Good to know the swap will work out of the box =).

I'll add in some SMTP details, and see how things go. The docs do briefly mention a "nightly" email (http://doc.freenas.org/9.3/freenas_system.html#email), however, I can't find anything about whether that period is configurable, whether it can only send alerts on error conditions instead of every night, or tweak what goes content goes into the email.

Hmm, so if RaidZ-1 with 4 drives is bad idea, what is a better setup with only 4 drives? Is mirroring the only option I have?

I believe there's some slightly hacky ways to fit a fifth SATA drive into the HP MicroServer (E.g. http://homeservershow.com/schoondoggy-gen8-drive-mount-hp-proliant-gen8-microserver-first-look.html). However, I will need to look into how feasible these are. If I am able to get five drives going, would RaidZ-2 be the most suitable setup for five drives then?

Finally, I'm sorry I didn't clarify the hardware properly. I had assumed the HP MicroServer's Gen8 were common enough that people would know it by name dropping...haha.

The host is a HP ProLiant MicroServer Generation 8 - the spec sheet is available at http://h20195.www2.hp.com/v2/GetDocument.aspx?docname=c04123182. This HP MicroServer has an in-built SATA controller card, a HP Smart Array B120i (http://h18013.www1.hp.com/products/...arraycontrollers/dynamicsmartarray/index.html). I am not using any of the hardware RAID features from the controller card - the disks just happen to be plugged in via this controller, and this is why I assume it was the controller card that was beeping when the disk went bad. I have 10GB of RAM in there. Two of the drives are Seagate Barracuda's 7200.14 (ST3000DM001-9YN166), and two of them are Hitachi Deskstar 5K3000 (HDS5C3030ALA630). I have now replaced one of the Seagates with a Hitachi Deskstar NAS 7200 (HDN724040ALE640).

Regards,
Victor

 

[SweetAndLow]

With 4 drives raidz2 use possible and you will get ~50% capacity. You should read the FreeNAS handbook cover to cover if you really want to know how things work. With emails you get scrub reports, smart test failure, reboots, security output. You only get an email if something is wrong, not nightly.