Workaround for DEFRAGMENT

[SR Rao]

One of the FS is around 37GB, and because of the application running and the load it takes, the defragment (defragment -n -v DOMAIN) fails with an error pertaining to the memory.
# defragment -n -v mds02
defragment: Gathering data for domain 'mds02'
defragment: Can't allocate memory for expanded xtnt desc table
defragment: Can't allocate memory for expanded xtnt desc table
defragment: Error = ENO_MORE_MEMORY (-1054)
************* PROGRAM ABORT **************

defragment: Error occurred during pass 1 on volume 4. Continuing...


defragment: chk_move_file()::Can't find file set hdr - tag 1.32769 - volume 1
defragment: Error occurred during pass 1 on volume 1. Continuing...
defragment: Can't allocate memory for analyzing storage bit map
defragment: Can't gather data for domain 'mds02'

If I manually copy the data from one FS to a new FS, will the data in the new FS still be fragmented???

Any other method of defragmenting???

[Victor Semaska_3]

SR Rao,

With regards to defragment, have you checked your ulimit? Maybe raising them, if possible, would help. Just a hunch but try raising data area with the -d option.

Vic

[Ivan Ferreira]

vfast is another method of defragmenting. Also, backup and restore is an old option. You should check your kernel memory parameters.

[Mark Poeschl_2]

Also, be aware that copying from one fileset to another WITHIN the same domain may do little good and may even make fragmentation worse depending on the shape of the free space in the domain. If you want to be sure of defragmentation during a copy or backup/restore the destination domain should have lots of contiguous free space.