Notification on disk fails

[lraymond]

Still under the n00b section as I am only 24 or so hours into my running device. I have setup 1 ZFS volume using 4 disks, and have an NFS setup using the 4 disks. I noticed when I first signed in a nice big red blinking ALERT saying I need to change the root password.

So I have my raidZ2 setup, exported as NFS and popped a drive out of the how swap bay. The monitor showed;
(ada1:ahcich1:0:0:0): lost device. I didn't get an email (email is setup and I did receive the test email), and no ALERT light blinking, etc. I took it up a notch and removed another;
(ada1:ahcich2:0:0:0): lost device. Again, no email, no alert, etc. and from my understanding this box should now be in degraded mode as I took all 4 2T disks, made 1 Volume (3.5T usable) on a ZFS volume.

When I click view disks, it only shows the 2, but when I click show volumes I see;
wf-zfs1 /mnt/wf-zfs1 209.0 KiB (0%) 3.5 TiB 3.5 TiB HEALTHY

Which to me is even more odd as shouldn't this show as degraded, or something?

* Update*
I also noticed the show disks had the 2 but when I clicked on volume information it showed all 4 which said online. When I re-inserted the disks, the show disks never updated (still only shows 2) and the show volumes still shows 4, I don't see a refresh list or something and also, there was no debugging at the console when the disk was re-inserted, so I don't have any way of telling the raidz2 status (so far it seems) or when a disk gets removed, or fails.

Thanks

 

[gpsguy]

What does section 1.4.4 of the manual say about this?

 

[cyberjock]

Sshhh. I'm sure he read the manual like a good boy. :)

 

[lraymond]

ok well @gpsguy, re-looking at 1.4.4 there is NO mention of mail/email/smtp/alert, etc. The only line that may help is "Until FreeBSD commits zfsd, its implementation of ZFS will not notice that a drive is gone until you reboot or put the volume on high load." but that really doesn't answer.

@cyberjock, sure, my 1st thing was let me read 258 pages on a computer screen before I ask a few simple questions.

Look, there is a difference between looking for answers, then asking for help (which I did plenty of things before I asked) vs a HELP ME in the forum title with NO real descriptions which I think I did.

So, with that, I did remove with NOTHING changing. I did put the disk under load and after a few minutes of read/writes, the alert light went yellow and blinked with an 'unknown' state. It didn't say it was running in degraded mode, but the disk status showed null which is fine since I didn't take it offline, I just removed it (trying to simulate a disk fail). The only negative thing I can see with that is I didn't get an email (YES, email notification is setup and I did get my test) so need to look around a bit more as to that, but alas ... you guys wait....

That server is going into the noc tomorrow so I can start to offload my dying EMC, and as I tune it more, I will know more and then be afriad! I am a HUGE forum advocate, and sarcasm is one of my biggest assets (ask my wife!) so I will be a cool Sr. member too one day and we will all look back and laugh at when I asked these questions! Now if I may be excused, I have another forum post to look at (unless one of you too answered that too!)

Seriously though, thanks guys ... have a great night!

 

[ProtoSD]

ok well @gpsguy, re-looking at 1.4.4 there is NO mention of mail/email/smtp/alert, etc. The only line that may help is "Until FreeBSD commits zfsd, its implementation of ZFS will not notice that a drive is gone until you reboot or put the volume on high load." but that really doesn't answer.

I can't count the number of times people have come here and tried/asked/posted pretty much the same thing you have about disk failure notifications, it's probably one of FreeNAS's biggest weak spots, but it's been covered over and over here, so you'll have to forgive us for being politely annoyed.
FreeNAS 9.1 is coming up soon, we'll probably see a beta in the next few weeks. That's doesn't mean we're going get zfsd right away, but it should be somewhere on the horizon, we've been waiting for it for probably more than a year. So for now, the kind of disk notifications that people are expecting are not working. That's why people like @Joeschmuck have put a bunch of work into the SMART email notifications as sort of an early warning system. So for now, backups, being prepared and understanding how things work (reading the docs) before you put your eggs all in one basket is the best thing to do. Like the other guys have said, I think you're really jumping the gun way too soon to be sending something to the NOC before you understand it better, It's probably going to bite you in the ass.

Anyway, welcome to the FreeNAS 8 forums! :D

 

[lraymond]

@Pro, thanks.

I did search a bit, and even the doc's are a bit fuzzy when it comes to the notifications, but I appreciate the direct feedback. I can understand and if that many people ask, may want a sticky on the n00b or even the FAQ as I do check both before I post. Thanks for the welcome also, I do love consumer forums as you get great feedback, meet interesting people, and honestly get a feel for how much the product is liked vs needed.

Now as for jumping the gun. I am kind of stuck due to hardware limitations, space and a failing device ($700 for an EMC harddrive is just ridiculous) so my thoughts are you can drive a car on a test track all day, but until you get on the real road, you don't know what's really going on. So, I have that 2T backed up safe, want to get this in and run some things parallel to both the EMC and the FN box and just switch the webservers to to read from that device. That will start to get real numbers, and I can then either post real questions confirming things are working, or post back some numbers with what was tried to improve, but until I get real traffic on, don't see a better way.

So thanks again for the welcome and looking forward to moving ahead with you guys :)

 

[Lucas Rey]

It's simple, for the disk notification failed, you have to make yourself a script that check the status aqnd send an email in case something goes wrong...

Just read here:
http://forums.freenas.org/showthread.php?1227-Send-email-from-shell
There's a way to send email and also the script for zpool check....