There is a workaround described in the forum (Re: ZFS issues? kernel: KLD zfs.ko: depends on opensolaris-not a)
I've tested this workaround, and it works... until the reboot :-(
After the reboot the zpool is gone and you have to import it again, or export it 'before' you want to reboot your NAS.
I'm really looking forward for a new build...
8 comments:
I just tried the 3953 version with ZFS and the before mentioned patch.
On both systems (vmware, real system (thinkpad)) there where no issues so far with ZFS. Pool created per GUI works fine and is online and ready after reboot.
So far I've done some stress testing with different datasets configured, including ZFS compression, have action on the datasets, simultanous read/write etc, and no corruption or panic sofar (checked with md5).
I know, both freenas and zfs are called experimental - but even with only 256MB ram that worked quite flawless for me.
Little difficult was the use of a swapfile and zfs with single disk configuration on the thinkpad. If time goes by and if it stays reliable and somebody is interested I may write some lines about the findings.
The zfs compression feature also works, tested gzip-1 and gzip-9 and lzjb. To use this you have to define a dataset. CPU consumption is significant but useable.
Now I will test your snapshot stuff, and hey, thanks for all that nice info :)
Regards, Stefan
Thank you Stefan, for this information. Looks like I have to doublecheck this :-)
Are you using an embedded installation?
You con use the opensolaris.ko from the FreeBSD live iso. ist works very well.
That is what I've done. After patching FreeNAS with the opensolaris.ko I was able to create a zpool. But after the reboot, the pool is not available and I have to import it by hand...
Harry, I use "full os" type installation (never tried the embedded one).
With my installs the zpool was always up and running after reboot.
I also disrupted power supply (by intention) whilst copying data to the box. Even after that the zpool was happy and running. All files that have been copied had correct md5. Status said no known errors - no fsck needed 8)
I will check that again with larger files and interrupt power just in middle of file so there is no chance to hold the already transmitted part in cache (thus box must have been written part of that file to the pool). I wonder what zfs does in case of interrupted file transfers....
Stefan, I've tested this with a full installation. And it works! Thanks for this info!
Harry, Great information as always... I have added a post about your discoveries here
Thanks, Gary
http://www.learnfreenas.com
Thanks Gary!
Post a Comment