Looking at the txg numbers, it's clear that labels on to devices that are unavailable now may be stale:
Krzys wrote: > When I do zdb on emcpower3a which seems to be ok from zpool perspective I get > the following output: > bash-3.00# zdb -lv /dev/dsk/emcpower3a > -------------------------------------------- > LABEL 0 > -------------------------------------------- > version=3 > name='mypool' > state=0 > txg=4367380 > pool_guid=4148251638983938048 > top_guid=9690155374174551757 > guid=9690155374174551757 > vdev_tree > type='disk' > id=2 > guid=9690155374174551757 > path='/dev/dsk/emcpower3a' > whole_disk=0 > metaslab_array=1813 > metaslab_shift=30 > ashift=9 > asize=134208815104 Here we have txg=4367380, but on other two devices (probably; at least on one of them) - txg=4367379: > But when I do zdb on emcpower0a which seems to be not that ok and get the > following output: > bash-3.00# zdb -lv /dev/dsk/emcpower0a > -------------------------------------------- > LABEL 0 > -------------------------------------------- > version=3 > name='mypool' > state=0 > txg=4367379 > pool_guid=4148251638983938048 > top_guid=14125143252243381576 > guid=14125143252243381576 > vdev_tree > type='disk' > id=0 > guid=14125143252243381576 > path='/dev/dsk/emcpower0a' > whole_disk=0 > metaslab_array=13 > metaslab_shift=29 > ashift=9 > asize=107365269504 > DTL=727 > > that also is the same for emcpower2a in my pool. What does 'zdb -uuu mypool' say? > Is there a way to be able to fix failed LABELs 2 and 3? I know you need 4 of > them, but is there a way to reconstruct them in any way? It looks like the problem is not that labels 2 and 3 are missing, but that labels 0 and 1 are stale > Or is my pool lost completely and I need to recreate it? > It would be off that reboot of a server could cause such disaster. There's Dirty Time Log object allocated for device with unreadable labels, and it means that device in question was not available for some time, so something weird might be going on with your storage a while back (prior to reboot)... > But I was unable to find anywhere where people would be able to > repair or recreate those LABELS. How would I recover my zpools? Any > help or suggestion is greatly appreciated. Have you seen this thread - http://www.opensolaris.org/jive/thread.jspa?messageID=220125 ? I think some of that experience may be applicable to this case as well Btw, what kind of Solaris are you running? wbr, victor _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
