Topic
  • 3 replies
  • Latest Post - ‏2008-12-23T08:04:01Z by tony.tang
tony.tang
tony.tang
9 Posts

Pinned topic can't mount gpfs filesytem,need help

‏2008-12-20T03:29:49Z |
I faced a gpfs problem on linux PC cluster ,The PC Cluster have four NSD server which have a direct connection to a san,and the gpfs version is v3.2.1-3 for linux x86_64.
Today somebody reboot the four NSD server ,and after rebooting the filesystem could not be mounted,I verified the disk using
mmlsnsd -m:
f2004548nsd09 C0A8648D4905C275 - f20045f (not found) server node
f2004548nsd09 C0A8648D4905C275 - f20046f (not found) server node
f2004548nsd10 C0A8648E4905C276 - f20046f (not found) server node
f2004548nsd10 C0A8648E4905C276 - f20047f (not found) server node
f2004548nsd11 C0A8648F4905C277 - f20047f (not found) server node
f2004548nsd11 C0A8648F4905C277 - f20048f (not found) server node
f2004548nsd12 C0A8648O4905C275 - f20048f (not found) server node
f2004548nsd12 C0A8648O4905C275 - f20045f (not found) server node

#mmlsdisk g2data15
Failed to read a file system descriptor
Input/output error
mmlsdisk:Command failed.Examine previous error messages to determine cause.
Following is the log on f20045f (one NSD server name) :
Fri Dec 19 13:24:05 CST 2008: mounting /dev/g2data15
Fri Dec 19 13:24:05.134 2008: Command: mount g2data15
Fri Dec 19 13:24:05.558 2008: I/O to NSD disk, f2004548nsd09, fails. No such NSD locally found.
Fri Dec 19 13:24:06.135 2008: Accepted and connected to 192.168.100.144 f20048f
Fri Dec 19 13:24:06.245 2008: I/O to NSD disk, f2004548nsd12, fails. No such NSD locally found.
Fri Dec 19 13:24:06.246 2008: Failed to open g2data15.
Fri Dec 19 13:24:06.247 2008: Cannot mount file system g2data15 because it does not have a manager.
Fri Dec 19 13:24:06.248 2008: The last file system manager was node 192.168.100.129 (f20033f). It has failed with error:
Fri Dec 19 13:24:06.247 2008: Input/output error
Fri Dec 19 13:24:06.248 2008: Input/output error
Fri Dec 19 13:24:06.247 2008: Command: err 5: mount g2data15
Fri Dec 19 13:24:06.248 2008: File System g2data15 unmounted by the system with return code 212 reason code 5
Fri Dec 19 13:24:06.247 2008: Input/output error
Fri Dec 19 13:24:06.248 2008: The current file system manager failed and no new manager will be appointed.
mount: /dev/g2data15: can't read superblock
Fri Dec 19 13:24:06 CST 2008: finished mounting /dev/g2data15

But every thing is OK in OS(Red Hat Enterprise Linux AS release 4 U6),I can find all luns and No any disk error reported,Is there any way to resolv this issue,There are some important data in this file system .
Thanks
Updated on 2008-12-23T08:04:01Z at 2008-12-23T08:04:01Z by tony.tang
  • janfrode
    janfrode
    20 Posts

    Re: can't mount gpfs filesytem,need help

    ‏2008-12-20T15:36:47Z  
    Not sure.. but I think I've gotten out of a similar situation simply by running "mmchmgr" to force a change of the filesystem manager.
  • esj
    esj
    104 Posts

    Re: can't mount gpfs filesytem,need help

    ‏2008-12-21T16:37:58Z  
    The "not found" in the mmlsnsd output usually means that either the disks do not show up in /proc/partitions, or that they do not have names that GPFS recognizes (i.e., names that start with sd, hd, vpath, emcpower), or that GPFS cannot open and read the disk.

    Since this thing was working, first verify that the disks show up in /proc/partitions
  • tony.tang
    tony.tang
    9 Posts

    Re: can't mount gpfs filesytem,need help

    ‏2008-12-23T08:04:01Z  
    Thanks for everyone,today I used od command to dump the diskarray:
    od -x /dev/sdh
    0000000 000000 000000 000000 000000 000000 000000 000000 000000
    *
    606500000 xxxxxx xxxxxx
    It means that the lun had been destoryed,perhaps somebody used dd command......
    It is a bad news.