very high inact_dirty value

From: hopehope_123 (hopehope_123_at_yahoo.com)
Date: 09/20/05


Date: 20 Sep 2005 07:31:14 -0700

Hi ,

The below is the output of meminfo of redhat linux 2.1 4 ia64 cpus ,
8gb.ram .

The value of inact_dirty is 2334352 kB and i have serious pagings (
sar -B shows heavy page in operations)

sar -B 1 10

Linux 2.4.18-e.37smp (tanidw1) 09/20/2005

04:28:47 PM pgpgin/s pgpgout/s activepg inadtypg inaclnpg inatarpg
04:28:48 PM 56746.00 4414.00 177877 146555 6925 66271
04:28:49 PM 55777.00 5502.00 177877 146555 6925 66271
04:28:50 PM 55371.00 3871.00 177877 146555 6925 66271
04:28:51 PM 53952.00 15.00 177877 146555 6925 66271
04:28:52 PM 40006.00 27646.00 177877 146555 6925 66271

[oracle@tanidw1 proc]$ cat meminfo
        total: used: free: shared: buffers: cached:
Mem: 8490827776 7987920896 502906880 0 200867840 3976101888
Swap: 25165725696 687030272 24478695424
MemTotal: 8291824 kB
MemFree: 491120 kB
MemShared: 0 kB
Buffers: 196160 kB
Cached: 3742224 kB
SwapCached: 140688 kB
Active: 2806032 kB
Inact_dirty: 2334352 kB
Inact_clean: 110880 kB
Inact_target: 1050240 kB
HighTotal: 0 kB
HighFree: 0 kB
LowTotal: 8291824 kB
LowFree: 491120 kB
SwapTotal: 24575904 kB
SwapFree: 23904976 kB
Committed_AS: 633471392 kB
HugePages_Total: 0
HugePages_Free: 0
Hugepagesize: 262144 kB

free

[oracle@tanidw1 proc]$ free
             total used free shared buffers
cached
Mem: 8291824 7838256 453568 0 198960
3751888
-/+ buffers/cache: 3887408 4404416
Swap: 24575904 670928 23904976

The inact_dirty pages must be written to the disk in order to be reused
, so why doesnt the operating system write these pages?

I have fund some messages similiar which says this can be a kernel bug
.

I will be appreciated if someone comments about this.

Kind Regards,
hope



Relevant Pages

  • Re: sorting a list and counting interchanges
    ... Unless I'm mistaken, this doesnt quite work, because it switches the ... parity of phase every time a comparison is made, ... a swap is made. ...
    (comp.lang.python)
  • Swapping table columns
    ... I got the following function to swap table columns from somewhere on ... the Internet ... integers representing the column numbers to swap. ... Problem is...it doesnt do anything. ...
    (comp.lang.javascript)
  • Re: sorting a list and counting interchanges
    ... >Unless I'm mistaken, this doesnt quite work, because it switches the ... >parity of phase every time a comparison is made, ... >a swap is made. ...
    (comp.lang.python)