For the people thinking about RAIDZ1

[snaptec]

Hello,
Everbody is talking about that raidz1 is dead and in case of a Drive failure another one could die during resilver.

An actual real world experience:

I still don't worry, as its just the backup from the backup

Feeled that i have to post this...
Greetz

Ps: that are just 1TB drives... Wd Enterprise class satas


Gesendet von iPhone mit Tapatalk

 

[Mirfster]

Hello,
Everbody is talking about that raidz1 is dead and in case of a Drive failure another one could die during resilver.

An actual real world experience:

Um, you are running RaidZ2.... Try that with RaidZ1 and let us know what your thoughts are on it then... :P

 

[Jailer]

Um, you are running RaidZ2.... Try that with RaidZ1 and let us know what your thoughts are on it then... :p

I think he's showing that in fact a drive can fail during resilver and that if it was RAIDZ it would be toast.

 

[gpsguy]

That's how I read his message.

I think he's showing that in fact a drive can fail during resilver and that if it was RAIDZ it would be toast.

 

[Mirfster]

I think he's showing that in fact a drive can fail during resilver and that if it was RAIDZ it would be toast.

That's how I read his message.

Fine be all logical and everything.. :P

My bad, if that was the case and apologies then. :)

 

[nojohnny101]

Thanks for posting.

Hopefully this post comes up in a google or forum search for someone contemplating raidz1 that doesn't make sense in their setup (more often then not the case).

 

[rs225]

The photo says nothing about raidz1 or raidz2. It tells us this person has a systemic problem that is causing corrupt data to be written or read from multiple drives, and not just a couple blocks, millions of blocks. That is the issue he needs to investigate, not post an image that validates his choice of parity on his single copy of data.(Yes, he says it is a backup...)

If you have a systemic problem, no amount of parity can be counted on to save your data. Raidz1 eliminates 99% of your drive or block error risk. raidz2 at best improves you a further 1%. Everything else is systemic risk, which you can only eliminate with your backup. This guy has posted an image of systemic risk. One of the purposes of scrubs is to reduce or reveal your systemic risk.

I am going to assume he is in the middle of a scrub, and not a resilver. If this is a scrub, you could have ALL of your drives degraded with checksum errors on raidz1 and not lose a single bit. In order to lose data, the data errors have to occur in the same block on multiple drives.

It is important to notice the errors are checksum, not read. When a drive can't read a block, it is a read error. When ZFS doesn't like what is read(which passed the drives checksum/ECC), it is a checksum error.

 

[snaptec]

Its a resilver. First Drive died, at 30% resilvering Another one died.
Resilvering with used hdds. It's Raidz2. And not the Backup, its the Backup from the Backup ;)


Gesendet von iPhone mit Tapatalk

 

[snaptec]

And yes its still degraded in cause of checksum errors.


Gesendet von iPhone mit Tapatalk