The problem may be instantiated in various forms; In case the quota file is corrupted, the vxrepquota(1m) command will show garbage values as limits.
In 6.0.3, 64 bit quota support was added which supported setting quota limits higher than 1TB. However, during rolling upgrade there is a window when two nodes could be on different patch levels - one node using 64 bit quota limit and other using 32 bit quota limit. When both the nodes do quota operations, one would update the external quota file based on 32 bit limits and the other would update it based on 64 bit limits, thus corrupting it.
It is advised that Rolling Upgrade should be performed only after quotas is turned off. A back-up of external quota file should be maintained to restore the original file in case quota file gets corrupted
To turn off quotas for a mounted file system:
vxquotaoff /mount_point
Please see man page vxquotaoff(1M)/vxquotaon(1M) for more information
To backup the quotas file:
cp /
Applies To
VxFS 6.0.3 on Linux and HPUX 11.31