George, Another option mentioned in the bug report is to ignore the resilvering progress by not using zpool status to see if it finishes (working around the zpool status bug).
If its not the zpool status bug that's causing this resilver to loop, then I would check the hardware, make sure the disk is seated properly, etc. Is the array degraded because the resilvering isn't finishing, or something else? Another h/w - s/w level to check (outside of ZFS) is whether this disk and disk label is visible with the format utility. You might describe your potential recovery scenario to the experts on zfs-discuss for more experienced feedback. Cindy ----- Original Message ----- From: "G. Mola" <[email protected]> Date: Monday, March 23, 2009 3:10 pm Subject: Re: [sysadmin-discuss] raidz1 looping silvering after hard drive To: [email protected] Cc: [email protected] > Hi Cindy, > > No problem at all.. I do appreciate all your help and I know how > difficult is it through email. > > Version: > SunOS solaris 5.11 snv_101b i86pc i386 i86pc > OpenSolaris 2008.11 with updates > > Resilvering seems to be completing (unless it dies at the last second, > when 0h00m left), but if I run zpool status maybe two minutes later, I > can see that resilvering had restarted (looks like probably > immediately and not after issuing "zpool status, although I don't have > a proof"). > > The bug ID you sent me seems to be different: running zpool status > does not restart ongoing resilvering from 0%. Resilvering in my case > restarts once no more time left and the progress is almost at 100% (I > actually never seen 100%, even though 0h00m was left, it was ninety > something percent.) > > I have not noticed any difference among different users, although I > have seen some postings that repeated resilvering occured if some > command was issued by root. > > Originally I did this: > geo...@solaris:~$ zpool clear distroclips c5d0 > cannot clear errors for c5d0: permission denied > > Then I switched to su to be able to run zpool clear. > > A short while after sending #zpool clear distroclips c5d0, I checked > zpool status and the array was still degraded and reselvering was in > progress again. > > What would happen if I manually stopped scrubbing, shut down, > disconnect c5d0, restart to let zfs complain (and hopefully forget > about c5d0s0/o). Then shut down, hook up an other drive to replace > current c5d0 and do a correct in-place replacement: > # zpool replace distroclips c5d0? > > > Thank you again Cindy. > > > > Hi George, > > > > Sorry for covering the basics again. Diagnosis by email is > > difficult. > > > > Which OpenSolaris release is this? > > > > Are you saying that the resilver never completes (possibly, > > really > > slow) or it restarts again when you run the zpool status > > command? > > > > I found this bug: > > > > http://bugs.opensolaris.org/view_bug.do?bug_id=6655927 > > zpool status causes a resilver or scrub to restart > > > > However, from your output it looks like the resilver is > > restarting > > whether you are e...@solaris $ or e...@solaris #. Is the > > e...@solaris $ > > a privileged user? > > > > Thanks, > > > > Cindy > > > > _______________________________________________ sysadmin-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/sysadmin-discuss
