Some non-recommend way, but if you really have no other way, you can try. 1) umount the MDT device 2) mount the MDT device as 'ldiskfs' 3) find the target 'progs' under /ROOT/ 4) backup the progs' lma EA 5) remove the progs' lma EA 6) rename 'progs' as 'tmpname' under the parent directory 7) remount the MDT device as 'lustre' 8) mount a new client, 'stat /absolute_path/tmpname'
If the 'stat' works, then you can rename 'tmpname' back as 'progs' under the 'lustre' mode. NOTE: Please backup the 'progs' lma EA before make any changes, then if something was wrong, you can restore the 'progs'. -- Lucky, Nasf > -----Original Message----- > From: Tommi T [mailto:tommi_...@yahoo.com] > Sent: Thursday, June 12, 2014 10:30 PM > To: Yong, Fan; lustre-discuss > Subject: Re: [Lustre-discuss] Directory access triggers OI scrub > > Yes, also with different client versions. > > el6: lustre-client-2.1.6/2.5.1 > > cray: lustre-client-2.4.1-3.0.101_0.29_default > > > > > On Thursday, June 12, 2014 5:18 PM, "Yong, Fan" <fan.y...@intel.com> > wrote: > Have you tried it with new mounted client after the upgrading? > > -- > Regards, > Nasf > > > > > > -----Original Message----- > > From: Tommi T [mailto:tommi_...@yahoo.com] > > Sent: Thursday, June 12, 2014 8:58 PM > > To: Yong, Fan; lustre-discuss > > Subject: Re: [Lustre-discuss] Directory access triggers OI scrub > > > > Hi, > > > > Still same error. I tried ls -ali /absolute/path/progs and > > ${parent}/progs > > > > Regards, > > > > Tommi > > > > > > > > > > On Thursday, June 12, 2014 6:01 AM, "Yong, Fan" <fan.y...@intel.com> > > wrote: > > Go to up layer directory, then try "ls -ail ${parent}/progs" to see > > what will happen. > > > > -- > > Best Regards, > > Nasf > > > > > > > > > > > -----Original Message----- > > > From: lustre-discuss-boun...@lists.lustre.org > > > [mailto:lustre-discuss- boun...@lists.lustre.org] On Behalf Of Tommi > > > T > > > Sent: Wednesday, June 11, 2014 5:11 PM > > > To: lustre-discuss > > > Subject: [Lustre-discuss] Directory access triggers OI scrub > > > > > > Hi, > > > > > > One user complained that he cannot access one directory. Accessing > > > that directory triggers OI scrubbing on the MDS but it does not find any > errors. > > > > > > Filesystem is created with lustre 2.1.3 tools and now server side is > > > upgraded to > > > b2_5 branch b7243c0 +kernel-2.6.32.431.17.1.el6_lustre.Is there some > > > way to debug this further to find out what the problem really is? > > > > > > > > > d????????? ? ? ? ? ? progs > > > > > > > > > ls -la progs > > > ls: cannot access progs: Operation now in progress > > > > > > > > > lfs getstripe progs > > > llapi_semantic_traverse: Failed to open 'progs': Operation now in > > > progress (115) > > > > > > > > > MDT side: > > > MDT0000: trigger OI scrub by RPC for [0x2012f90d5:0x6641:0x0], rc = > > > 0 [1] > > > > > > stat /homeappl/.lustre/fid/[0x2012f90d5:0x6641:0x0] > > > stat: cannot stat `/homeappl/.lustre/fid/[0x2012f90d5:0x6641:0x0]': > > > Operation now in progress > > > > > > > > > name: OI_scrub > > > magic: 0x4c5fd252 > > > oi_files: 1 > > > status: completed > > > flags: > > > param: > > > time_since_last_completed: 889 seconds > > > time_since_latest_start: 991 seconds > > > time_since_last_checkpoint: 889 seconds > > > latest_start_position: 12 > > > last_checkpoint_position: 268435457 > > > first_failure_position: N/A > > > checked: 18208728 > > > updated: 0 > > > failed: 0 > > > prior_updated: 0 > > > noscrub: 1586 > > > igif: 80 > > > success_count: 18 > > > run_time: 102 seconds > > > average_speed: 178516 objects/sec > > > real-time_speed: N/A > > > current_position: N/A > > > lf_scanned: 0 > > > lf_reparied: 0 > > > lf_failed: 0 > > > _______________________________________________ > > > Lustre-discuss mailing list > > > Lustre-discuss@lists.lustre.org > > > http://lists.lustre.org/mailman/listinfo/lustre-discuss _______________________________________________ Lustre-discuss mailing list Lustre-discuss@lists.lustre.org http://lists.lustre.org/mailman/listinfo/lustre-discuss