[zfs-discuss] Expanding pool

Sander Klein roedie at roedie.nl
Thu Oct 31 08:10:42 EDT 2013


Hey Gordan,

On 31.10.2013 12:10, Gordan Bobic wrote:
> Why not simply replace one disk at a time?

Well, the server is not really near my location. So, if I have to drive 
there every time to replace one disk. Replacing 2 at a time saves me 
time to drive there. It will most likely not speed up the process.

> Also, does RAIDZ2 give you enough redundancy for your new disk sizes?
> The expectation that you will get at least one onrecoverable sector
> for every 5TB transferred is probably a good place to start assessing
> this.

It will become an array of 10 4TB disks. Which is not my favorite 
setup, but then again, it only has to run like this for a couple of 
months.

I might be wrong, but if I do a scrub before starting the replace with 
2 disks, the changes of encountering a checksum error during the rebuild 
are pretty slim. The data is backupped so in case of total disaster I 
can recover again.

The thing is, I could do just recreate the pool with the new disks, and 
recover everything from backup. But that will cost me 24 hours of 
downtime. I'm willing to take a chance. I might be lucky, or I might 
have my downtime anyway.....

Greets,

Sander

To unsubscribe from this group and stop receiving emails from it, send an email to zfs-discuss+unsubscribe at zfsonlinux.org.



More information about the zfs-discuss mailing list