New answers tagged zfs
Start with disabling atime if not already done. You might also investigate setting primarycache=metadata impact.
The ZFS feature set copies=2 was not designed to overcome disk failures (see the comments on the following Q&A): ServerFault: How do you restore the correct number of copies after losing a drive The issue here being that as your pool is made of single disks, ALL disks must be healthy for import. This can only be done in your case by recovering ...
Could you run your IO heavy tasks from yet-another-VM? There will be a lot of overhead, yes, and you still have to make sure that you don't overload your NFS, but that way all traffic is treated equally.
Consider QuantaStor. They are experienced with Gluster atop ZFS (and the 30% performance hit). Ask them about snapshotting strategies.
Just to be mentioned, if you go with RAID-Z, you may not easily "add some more drives" for any of RAID-Z.
I'd suggest dedicated drives for OS and dedicated disks for data. NexentaStor is software RAID and the loss of an OS disk can be painful. You can combine L2ARC and ZIL, but the class of SSDs you're probably using won't be helpful as ZIL devices. I understand the desire to try to use all of these ZFS features, but you really have to do it the right way. I ...
Good news is you can change the vdev configuration scheme by exporting and re-importing your pool. (from ZFS on Linux docs) admin@Test-ZFS:~$ sudo zpool status tankz pool: tankz state: ONLINE scan: resilvered 15K in 0h0m with 0 errors on Tue Feb 25 16:36:18 2014 config: NAME STATE READ WRITE CKSUM tankz ONLINE 0 ...
It looks like the system did not handle a pulled disk very well. Although most of it seemed to work correctly, the format and zpool create commands hung even after the missing disk has been re-inserted. Rebooting the system helped matters - a fast reboot was sufficient.
Scrub your pool again (if you haven't already): zpool scrub zstorage That error is telling you that inode <0x9f115> is corrupt (deleting the file broke the filename->inode mapping, so it's just reporting the inode now). Either something still has the file open or the metadata just needs to be cleaned up (which a scrub should do). To clear the error if ...
My recommendations vary a bit based on the distribution and approach to ZFS... SSD-only pools work well - However, you need to think very carefully about the controller, backplane and drive layout. Main considerations are: no expanders, use pure SAS HBAs, ZFS mirrors are king. 15k SAS disks in 2014 are Meh - If you need the random I/O performance of SSDs. ...
It looks like your zpool.cache file is corrupted (I had a few of those in the early FreeBSD ZFS days). So as soon as the system try to access your pool, it stay stuck while trying to read its cached info. Simply: boot as "single user" and do not launch or access your zfs pool go to the directory where is your zpool.cache file is (default is /boot/zfs/) ...
Set autoreplace=on for your pool and use like or similar disks. Resilvering occurs automatically when that flag is set on the pool. If a hot-spare is defined in the pool, it will also rebuild automatically if autoreplace is on. There's nothing more to really consider.
The autoreplace ZFS pool property is exactly that. From the ZFS Administration Guide, about the : If set to on, any new device found in the same physical location as a device that previously belonged to the pool is automatically formatted and replaced. Also beware, if physically replacing the faulty disk with a new properly working one will trigger ...
Hmms, look like the system (Nexenta) is getting too many NFS connections or you're hitting a bug. You could try to up the maxdupreqs, that could be masking the issue, but let's try that first: echo 'rpcmod`maxdupreqs/W0t4096' | mdb -kw echo 'rpcmod`cotsmaxdupreqs/W0t4096' | mdb -kw To check what it's current value is use: echo ...
Top 50 recent answers are included