JFS 3.5 inode cache question

[likid0]

Hy people,

Looking at the vxfs stats, i have a question:

vxfsstat -v / | grep ino
vxi_iflush_cut 0 vxi_icache_allocedino 573685
vxi_icache_freedino 0 vxi_icache_curino 573685
vxi_icache_inuseino 1981 vxi_icache_maxino 572770
vxi_icache_peakino 573685

is it normal to have the peak well over a thousand of your vxi_icache_maxino ?

I have read that its normal to see 40 - 50 over the maximun, but a thousand?

does this mean i should increase the vx_ninode ?

Thnx

[Steven E. Protter]

Shalom,

Its a sign of heavy use.

Do you have OnlineJFS? Do you defragment.

I would not consider making changes unless you have a definite complaint from the users of this filesystem.

SEP

[likid0]

Yes, this application is heavy on IO,we defragment once a week, but even then some fs,are very fragmentated, like this ones:

Extent Fragmentation Report
Total Average Average Total
Files File Blks # Extents Free Blks
4096588 4 1 58016378
blocks used for indirects: 609144
% Free blocks in extents smaller than 64 blks: 16.16
% Free blocks in extents smaller than 8 blks: 16.16
% blks allocated to extents 64 blks or larger: 0.49


We are having some problems, with the application,we are triying to exclude things