Freenas 8.3 Beta1 not detecting faulty/disconnected disks

[mephisto]

the only problem with solaris is the lack of GUI for ZFS, everything to be done properly is command line. Just nappit is a bit messy in my opinion.

I would not leave my data hoping for the best with freenas to find out a disk failed and several months later I lost all my data because the hot spare didn't kick in. Physically checking drives goes against the purpose of the whole thing.

 

[cyberjock]

I have an audible alarm on my controller so that's one way I'd find out a drive has failed within seconds ;)

Also, the nightly emails that FreeNAS sends out a status and if you look at them they will say "DEGRADED" for the zpool.

 

[mephisto]

I have an audible alarm on my controller so that's one way I'd find out a drive has failed within seconds ;)

Also, the nightly emails that FreeNAS sends out a status and if you look at them they will say "DEGRADED" for the zpool.

I thought that the reporting problem was also down to the system. Last time I used freenas was version 3 I think so I was sort of not on pair with what happened on the last 3 years

 

[cyberjock]

When I did a test in another thread the nightly email showed DEGRADED for the zpool, but the GUI green light was still lit. /shrug

Overall, I'm going to keep checking the emails every day, but I'll definitely rely on that annoying loud alarm on the controllers. You can hear them across the house and through 2 closed doors! When my server was in the basement I could hear the alarm in my garage. LOL. There's no way I'll not notice something like that.

This test was done this month using an 8.3 nightly build.

Edit: Changed the first sentence from "it" to "the nightly email".

 

[mephisto]

When I did a test in another thread the nightly email showed DEGRADED for the zpool, but the GUI green light was still lit. /shrug

Overall, I'm going to keep checking the emails every day, but I'll definitely rely on that annoying loud alarm on the controllers. You can hear them across the house and through 2 closed doors! When my server was in the basement I could hear the alarm in my garage. LOL. There's no way I'll not notice something like that.

This test was done this month using an 8.3 nightly build.

Edit: Changed the first sentence from "it" to "the nightly email".

Does the email alarm work fine for a instant alert like when the drive failed or it works just as a regular email every x hours?

My concern is, the drive fails, I don't get any notification until X days and then by the time I replace the drive another one fails.

 

[cyberjock]

Does the email alarm work fine for a instant alert like when the drive failed or it works just as a regular email every x hours?

My concern is, the drive fails, I don't get any notification until X days and then by the time I replace the drive another one fails.

You get the nightly email every night. For me I have it sent to my phone as a text message. I typically receive it about 3am. You don't get an immediate email when the drive fails, but the next nightly email will warn you of the degraded status of the zpool.

Personally, I don't do RAIDZ at all. I only do RAIDZ2(and soon RAIDZ3 will be an option). Additionally I always keep a spare drive in the box ready to replace any failed drive. I don't have the ability to go buy a new hard drive since I live in a small town of less than 9k. I'd have to take an all-day trip to buy a new hard drive or order one online. Since I don't want to wait I keep one on standby. I feel that the daily email plus having a spare hard drive on standby for a RAIDZ2 is plenty of safety margin.

 

[mephisto]

You get the nightly email every night. For me I have it sent to my phone as a text message. I typically receive it about 3am. You don't get an immediate email when the drive fails, but the next nightly email will warn you of the degraded status of the zpool.

Personally, I don't do RAIDZ at all. I only do RAIDZ2(and soon RAIDZ3 will be an option). Additionally I always keep a spare drive in the box ready to replace any failed drive. I don't have the ability to go buy a new hard drive since I live in a small town of less than 9k. I'd have to take an all-day trip to buy a new hard drive or order one online. Since I don't want to wait I keep one on standby. I feel that the daily email plus having a spare hard drive on standby for a RAIDZ2 is plenty of safety margin.

I don't wanna use Z2 because of the overhead dual parity will cause, I rather prefer to have a spare one for now. How the hot spare work? Does it kick in properly or not as the GUI displays?

 

[cyberjock]

I don't use hot spares so I don't know. I prefer to have a brand new drive to use instead of having a hot spare sitting there running for months.

 

[mephisto]

I was wondering about nas4free, it is based on freebsd9.

Will give it a go later on and check how thing go, zfs verison 28 as well is a bonus

 

[yaneurabeya]

You get the nightly email every night. For me I have it sent to my phone as a text message. I typically receive it about 3am. You don't get an immediate email when the drive fails, but the next nightly email will warn you of the degraded status of the zpool.

Personally, I don't do RAIDZ at all. I only do RAIDZ2(and soon RAIDZ3 will be an option). Additionally I always keep a spare drive in the box ready to replace any failed drive. I don't have the ability to go buy a new hard drive since I live in a small town of less than 9k. I'd have to take an all-day trip to buy a new hard drive or order one online. Since I don't want to wait I keep one on standby. I feel that the daily email plus having a spare hard drive on standby for a RAIDZ2 is plenty of safety margin.

Except you must pay closer attention to performance degradation and/or "hotspotting". The only thing that does a reasonable job at making drives fall out in ZFS (pre-zfsd) is a scrub, and that's assuming your driver is doing the right thing talking to the entire storage stack (mfi didn't do that in 9.0 and mps didn't do that prior to the LSI supported driver change, and gmultipath didn't do that until recently).

In all cases, test before assuming that the driver/storage stack will do what you hope it will do.