Each night, as part of the backup script, we copy some large database files from a “live” filesystem to a “copy” filesystem. We then backup (using “tar”) from the “copy” filesystem. This is so the backups can continue whilst we reactivate access to the live databases.
This system has worked perfectly well since its inception about 18 months ago.
About 2 weeks ago, we noticed that the backup script was “tar”’ing a 0 byte file from the “copy” filesystem. On further investigation we could see that the filesystem was being reported as full, yet it only held about half of its capacity in data.
More to the point (and ensuring it wasn’t in use and nobody else was on the system) we tried to unmount this filesystem , only to receive the message that it was in use.
A system reboot seemed to resolve the problem, allowing us to unmount/mount the “copy” filesystem and it would also show a correct figure on how full it was.
The problem returned however after 2 days though.
Today, we have rebooted the system, unmounted this filesystem and ran an fsck on it. No problems were reported (and therefore none fixed!). Similarly we ran the same command on the originating unmounted “live” system with the same results.
We’ve ran “diag” on the underlying physical disks and all is fine.
We are still not clear as to what could be causing this problem. Any ideas
The system is an IBM RS6000 F80, about 2 years old, running AIX 4.3
This system has worked perfectly well since its inception about 18 months ago.
About 2 weeks ago, we noticed that the backup script was “tar”’ing a 0 byte file from the “copy” filesystem. On further investigation we could see that the filesystem was being reported as full, yet it only held about half of its capacity in data.
More to the point (and ensuring it wasn’t in use and nobody else was on the system) we tried to unmount this filesystem , only to receive the message that it was in use.
A system reboot seemed to resolve the problem, allowing us to unmount/mount the “copy” filesystem and it would also show a correct figure on how full it was.
The problem returned however after 2 days though.
Today, we have rebooted the system, unmounted this filesystem and ran an fsck on it. No problems were reported (and therefore none fixed!). Similarly we ran the same command on the originating unmounted “live” system with the same results.
We’ve ran “diag” on the underlying physical disks and all is fine.
We are still not clear as to what could be causing this problem. Any ideas
The system is an IBM RS6000 F80, about 2 years old, running AIX 4.3