The Great Southern California Blackout....

[ProtoSD]

Well I'm sitting in my cozy little 'hut' in the mountains almost 1,000 miles away, but my NAS is sitting power deprived in San Diego. I didn't find out about the blackout until long after my UPS died, and I'm running beta-4, so it would have only mattered if I had found out soon enough to login and manually shut down. The router and modem are on another UPS and last I checked I could still get in.

While I'm pretty confident ZFS will come through for me (last scrub was 19 days ago), I'm curious to hear about anyone else's experiences with power outages and ZFS. I've heard of some people having problems, while others had none.

Durk, how're things going for you there?

 

[Durkatlon]

Yeah, we've been without power for about 7 hours now. I shut down my 4 FreeNAS boxes using the 4-second power button salute. Not worried about the filesystem. This is why I got rid of my XFS based commercial NASes.

All radiostations are relaying KOGO which is doing continuous outage related coverage. Last we heard it could be another 24 to 48 hours for power to come back.

Shutting off my phone now to conserve what charge I have left. Been an interesting experience! :D

 

[louisk]

I did a bunch of tests when I was originally deploying zfs. Power pulls, drive pulls, everything I could think of. I never had an issue with data on disk being corrupted. This was ~3yrs ago, circa FreeBSD 7.0, when ZFS was still shiny and new. I've been using it ever since (on various systems), and for as long as FreeNAS has offered it.

 

[Milhouse]

I think with a system like ZFS the more likely problem with sudden and unexpected power loss is the potential physical harm that can result once power is restored.

I once worked for a company with a server farm of over 1000 servers and while the primary PDU was being serviced the backup PDU failed unexpectedly and almost 5% of servers when rebooted had some form of physical failure (either motherboard or hard disk related) and a smaller number of systems had to be scrapped entirely (very old kit, probably only kept working because they had never been powered off in years).

 

[ProtoSD]

@Milhouse - I think that your story is a good argument for raidz2. It seems like even when you think you're covered with some type of secondary backup, that's when it's likely to fail.

My NAS seems to have recovered ok, I'll know for certain after I run a scrub tonight which takes about 14 hours.

Update: Scrub finished successfully, my beta-4 system survived the power failure and came back up on its own without any problems.