[OpenIndiana-discuss] Diagnosis help needed

[Jan Owoc]

On Sun, Jun 24, 2012 at 1:01 AM, michelle <michelle at msknight.com> wrote:
> Situation - home-type standard PC, 4gig of RAM, running two SSDs in a
> mirrored root raid pool. Three 2tb hard drives in a raidz.
>
> System is...
>
>             OpenIndiana Development oi_151.1.4 X86 (powered by illumos)
[...]
>
> I have an external, "toaster" which takes two hard drives, one is connected
> via e-sata; the other is running via USB (although for this instance, there
> was no drive in teh socket) because I've had a long running battle to try
> and get an affordable (to me) e-sata card that will give me another e-sata
> channel.

So it's a 2-bay RAID enclosure with either USB or eSATA connections.
One of the two bays are occupied, and how is the enclosure being
connected?


> The ZFS set was getting full; something like only 50gig free.

You probably have two zpools - one is the mirrored "rpool", while the
other is your data pool, say, "tank". Am I understanding that it's
"tank" that has 50 GB (out of 4TB) free, while "rpool" does not have
any problems?


> I was starting
> file copies off the server to an external drive via an SMB client, and going
> to bed, to wake up and find the process had frozen. Diagnosis led me to the
> server, which appeared to hang on any log on attempt. It even didn't listen
> to the power button properly.

Is this external drive the RAID enclosure discussed above, that you
connected via Ethernet and it shows up as an SMB device, or is this on
a separate computer?


My thoughts are:

1) if "rpool" is not full, then the system should not freeze.
Depending on any snapshots, even removing files from "tank" may fail
and if you do it via SMB, as opposed to over a local command line, you
won't know why. Could you try logging on to the system, and copying
the files from the server to a client, so you see any local error
messages on the command line?

2) if, for some reason, data loss crept in, ZFS will refuse to return
bad data. Could you run a "zpool scrub" on each of the pools and then
verify that they are healthy?


> Could the lack of free space on the ZFS set also have caused a problem, or
> is it likely that the weight of another problem, possibly the USB external
> drive connection, caused it to keel over?

Not sure if it will help, but could you give details on what the
enclosure is (brand/model), and what motherboard/USB controller are on
the server?


Jan