On Wed, Jul 03, 2013 at 08:35:48PM +0200, Torbjørn wrote: > Hi btrfs devs, > > I have a btrfs raid10 array consisting of 2TB drives. > > I added a new drive to the array, then balanced. > The balance failed after ~50GB was moved to the new drive. > The balance fixed lots of errors according to dmesg. > > Server rebooted > > The newly added drive were no longer detected as a btrfs disk. > The array was then mounted -o recovery > I ran btrfs dev del missing, and everything seemed to be fine. > > After this I ran a scrub on the array. > The scrub was soon stopped by the oom-killer. > > After another reboot I started a new scrub. > About 3TB into the scrub over 10 GB of memory was being consumed. > The scrub had then fixed roughly 3,000,000 errors. > > Canceling the scrub and resuming it frees the 10 GB of memory.
Thanks for the report. This looks like the same problem that was fixed by https://patchwork.kernel.org/patch/2697501/ Btrfs: free csums when we're done scrubbing an extent but I don't see it included in the current for-linus branch. We want this in the 3.10.x stable series and according to stable tree policy it has to be merged into Linus' tree first. david -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
