Openindiana napp-it log for disk error

[RandyV]

Hi all,

1> Does anybody know if there is a logfile in openindiana where you can read about when a disk broke down? I use napp-it en see that a disk in an rpool mirror is now unavailable. Ofcource disk has been replaced but I would like to be able to trace when it happened.

2> Is it possible to place a spare into the rpool so that any broken disk in the rpool gets replaced automatically, or can it just not be done because of grub stuff.

3> when a spare is placed into any zpool, is it not so that the os should automatically replace a broken disk with the spare? If so is there anything that needs to be done after that (like replacing the broken disk and then putting it back into the pool by replacing it with the spare now in use...). What would be best practise in such a case?



Rgds,

Randy

 

[danswartz]

1. I *think* you can look at /var/adm/messages.
2, The grub stuff would be the catch. However, you could install grub on that spare ahead of time, but if you were going to do all that, why not just put it in the rpool to begin with. Also, an rpool spare makes no sense unless the rpool is already mirrored. I wouldn't bother with double-redundancy on the rpool.
3. I don't use spares, so I can't comment. I seem to recall there is a pool property that says whether to autoreplace a bad drive.

 

[RandyV]

Thanks for your respons Dan.
That's the thing, /var/adm/messages shows nothing and neither does the syslog. So maybe someone can tell me which logfile is used if any. I also would appreciate it if anybody can shed some light on the third question.

 

[_Gea]

If you only need the time of a failure, you can setup an alert job and you get email one minute after a failure.

about 3.
if you add a spare to a pool, its state is available.In case of a disk failure it replaces the faulted disk
(spare is inUse).You must then replace the faulted disk and do a napp-it disk-replace (active spare with the new disk).
The spare is then available again.

 

[RandyV]

Thanks for your response Gea. This clarifies things.

 

[RandyV]

Would it be an option to put the rpool in a hardware mirror via the hardware raidcontroller. I know that with zfs you are not supposed to use hardware raid options. What would be the pitfall in this case (putting all other discs in passthrough mode so that zfs can handle them directly)

 

[_Gea]

Would it be an option to put the rpool in a hardware mirror via the hardware raidcontroller. I know that with zfs you are not supposed to use hardware raid options. What would be the pitfall in this case (putting all other discs in passthrough mode so that zfs can handle them directly)

You have always the same problem with hardware-Raid
- if a disk fails completely, a hardware Raid is fine
- if there are errors that can only be detected via checksums, ZFS will detect the error
but the hardware Raid thinks that everything is ok

-this results in a disk (hardwareraid) failure from ZFS view (happens more often than a complete disk failure)
- on a software-Raid, ZFS is able the detect which mirrorpart is good and repair the faulted data)
(ZFS self-healing feature, not available with hardware-Raid)

It you look for a ZFS solution:
http://constantin.glez.de/blog/2011/03/how-set-zfs-root-pool-mirror-oracle-solaris-11-express


What I do with my bootable hardware Raids to avoid the checksum problem:
- add the second disk to build the mirror
- remove the second disk to have a bootable cold backup
- on systemchanges, plug it in to rebuild and remove afterwards

 

[RandyV]

Thank you for this information!