Extending a JFS filesystem on-line with an active database.

[Sheila McKenzie]

Have a database on-top of a JFS filesystem. Want to extend the filesystem using On-line JFS and the "fsadm" command just "hangs" -- sits there -- gave up after several hours. If the database is brought down, the on-line jfs command works.

Thoughts???

[RAC_1]

How exactly you are doing it. Can you post the command lines you used?

Is Online JFS installed properly? Check with swverify.
It should be like this.

lvextend -L xxM /dev/vgxx/lvolxx
fsadm -F "fstype" -b xxxxM /mount_pointof_file_system.

Do you have tusc tool? Is there any error in syslog? You can start it with tusc and see where it hangs.

[Geoff Wild]

How full is the filesystem? If 100% - then it won't work...

For example:

lvextend -L 11264 /dev/vg18/lvAPPL
fsadm -F vxfs -b 11264M /export/APPL

Rgds...Geoff

[Victor BERRIDGE]

Hi,
2 things:
1) A good practice would be defragmenting the filesystem before such manipulations:
fsadm -F vxfs -D -d -E -e -s $FILESYSTEM
2) As already mentioned fsadm -b needs extra space prior to the expansion for the extra growth of its structural files.

All the best
Victor

[Marco Santerre]

I woudl certainly look more in the patch area or in the fact that the product is installed correctly.

If a file system would have been full at 100%, it's true that the command fsadm would not have worked but it would nopt have hung.. it would give you an error message saying it cannot do the command.

[Victor BERRIDGE]

Good remark Marco, I missed the "hangs"...
One thing missing also is what version of OS?
10.20 , 11.00 or 11.11?

Victor

[Steven E. Protter]

Its true if the filesystem is too full it might have trouble, but OnlineJFS should work.

This is probably an installation error or some process has the device for the filesysetm on hold.

At this point you might want to down the database and run fusers -cu /fsname

Most likely problem here is patching or corrupt installation.

SEP

[Colin Topliss]

Checking your patch level is always a good thing to do.

The only other information we're missing (aside from the OS) is the type of disks you are using. Are they internal, SCSI array, or fibre?

I'd also want to check how busy the disks are that your database sits on. If they are busy there would be a delay (I certainly wouldn't expect hours though).

Have you applied patches recently? Is this the first time you have had to extend the filesystem in this manner, or has it been done before with no problems?

Were there any error messages (such as I/O errors being reported - look at dmesg and syslog). Its possible you are getting SCSI resets if the I/O subsystem is hanging (depending on your architecture of course).

[Sheila McKenzie]

Thank-you all for your responses. You are confirming that this should work.

It is strange that when the database is brought down, the exact commands to extend the filesystem that were used when it was up now works.

Will look into bring system up to the latest patches.

Thanks again.