fsadm error

[moliere francois]

Hi!
I have having a little problem rying to figure out following error by fsadm:
fsadm -F vxfs -b 1024000 /usr
vxfs fsadm: /dev/vg00/rlvol8 is currently 1634304 sectors - size will be reduced
vxfs fsadm: allocations found in shrink range, moving data
vxfs fsadm: cannot shrink /dev/vg00/rlvol8 - blocks are currently in use.

I don't to unmount /usr to srhink it, so any help on working around this problem is appreciated.
M

[Pete Randall]

Did you re-org it first?

fsadm -F vxfs -d -D -e -E /usr


Pete

[James R. Ferguson]

Hi:

While degragmentation *may* allow a subsequent attempt to shrink the filesystem size to work, the failure rate is high for JFS versions prior to 3.3.

If you *must* shink the filesystem, you could backup the data; boot into single user mode; 'lvreduce' and 'newfs' the filesystem to the intended size; and then reload your data.

Another alternative, since this is a vg00 filesystem, is to re-Ignite your server using an Ignite recovery tape.

Regards!

...JRF...

[S.K. Chan]

As mentioned by JRF, FS shrinking operation is best done by backing it up and recreating the LV. In you case (if you have OnlineJFS) the command you use should work. If you don't hv OnlineJFS the only way is to do this in single user mode.
# fsadm -F vxfs -b 10240000 /dev/vg00/rlvol8
and then resize the LV accordingly with lvreduce.

[Michael Steele_2 · Accepted solution]

This error seems to indicate that you don't have online JFS but only the base JFS version.

Does this work?:

fsadm -F vxfs -d -D -e -E /usr

If this fails then you will have to umount /usr in order to grow /usr but WILL NOT BE ABLE TO SHRINK IT safely without a reinstall.

[James R. Ferguson]

Hi (again):

Since it is the '/usr' filesystem tthat you want to shrink, obtaining a quiescent filesystesm would require a boot-up into single user mode where '/usr' isn't mounted.

Since the 'fsadm' binary lives in '/usr' with no version available in '/sbin' (AFAIK) you will need to mount '/usr' while still in single-user mode. *Then* you might (or not) be sucessful in dynamically shrinking the filesystem.

Regards!

...JRF...

[moliere francois]

The defrag. process of running fsadm -F vxfs -d -D -e -E did not solve the problem. However, I get a different error message now when I try to shrink /usr.
fsadm -F vxfs -b 10240000 /usr
vxfs fsadm: /dev/vg00/rlvol8 is currently 1634304 sectors - size will be increased
vxfs fsadm: write failure at block 14434302 : No such device or address.

I have booted in single user mode to reduce /usr, but the binaries reside in /usr as previously pointed out. Running fsadm on /usr in single user mode produced the following error:

fsadm: you don't have a license to run this program

I understand that I may not have the full version of online jfs since swlist is reporting:
B3929CA B.11.11 HP OnLineJFS
I am thinking that perhaps, I need to have the Advanced Vxfs FileSystem loaded.

[John Dvorchak]

First of all the second time you ran the fsadm command, you increased the size by 10x over the first time you ran the command. That is why you are getting the error that it can't EXPAND past block xxxx.
If you cut and pasted the commands as they were typed, then you didn't reduce it to 1gb you expanded it to 10.

[S.K. Chan]

Thanks John ... it's my typo mistake, extra "0". Check if the vxadv driver is in your kernel or not (you can use SAM) or
# grep vxadv /stand/system
OnlineJFS needs that driver to work.

[Michael Steele_2]

You Online JFS 3.3 license is stored and configured in /etc/inittab which is not read until run level 2.

I would not proceed with this reduction. You take the chance of truncating some of your O/S. This is because /usr is not contiguous. The entire lvol is utilized, hence the need to defragment. Although decrementing approaches contiguousness, it doesn't guarantee it.

One final note, you're defrag will probably fail if you're over 85% utilized. Some temporary work space is needed for the procedure.

[Michael Steele_2]

Regarding: "...Although decrementing ..."

...should read "...Although defragmenting...".

[moliere francois]

Ok, this is what I had to resort to in order to reduce /usr since nothing else worked.
created a 1 GB logical volume (lvol9) and mount it at /usr2
then copied the content of /usr to /usr2
changed /etc/fstab to reflect lvol9 mounting at /usr
rebooted
lreduce /dev/vg00/lvol8 -L 1024M
(note) I could not fsck lvol8 no matter what options I used. Apparently lvol8 was beyond repair and I had to newfs it instead.
newfs -F vxfs /dev/vg00/rlvol8
mount /dev/vg00/lvol8 /us2
restored the content from temporary /usr to /usr2
cp -R -p * /usr2
changed /etc/fstab to reflect lvol 8 mounting at /usr
reboot and it worked like a charm.
/usr has been reduced to 1 GB as needed.

Thanks to all who have responded. You guys are great!