Hi Patrick,

Thank you for responding so quickly. I took your
advice and ran the command you suggested. It seems
that inode_cache seems to continuously grow... as I
watch this command. dentry_cache follows too, but at a
much slower rate. What does this indicate?

Thanks,
- D.

Here is the output from one instance:


Every 1s: sort -nr -k 2 /proc/slabinfo | head -20     
           Sun Mar 17 03:34:45 2002

dentry_cache       46680  46680    128 1556 1556    1
:  252  126
inode_cache        45560  45560    480 5695 5695    1
:  124   62
buffer_head         8714   8840     96  219  221    1
:  252  126
size-128            1440   1440    128   48   48    1
:  252  126
size-32              904    904     32    8    8    1
:  252  126
vm_area_struct       634    760     96   19   19    1
:  252  126
filp                 360    360    128   12   12    1
:  252  126
skbuff_head_cache    336    336    160   14   14    1
:  252  126
blkdev_requests      280    280     96    7    7    1
:  252  126
cdev_cache           228    354     64    4    6    1
:  252  126
size-256             210    210    256   14   14    1
:  252  126
size-64              177    177     64    3    3    1
:  252  126
size-2048            152    152   2048   76   76    1
:   60   30
devfsd_event         126    169     20    1    1    1
:  252  126
ip_dst_cache         120    120    192    6    6    1
:  252  126
uid_cache            113    113     32    1    1    1
:  252  126
tcp_bind_bucket      113    113     32    1    1    1
:  252  126
size-32(DMA)         113    113     32    1    1    1
:  252  126
ip_fib_hash          113    113     32    1    1    1
:  252  126
size-1024            108    108   1024   27   27    1
:  124   62

--- Patrick Schaaf <[EMAIL PROTECTED]> wrote:
> On Sun, Mar 17, 2002 at 02:32:18AM -0800, Designer
> Seven wrote:
> > 
> > I'm running a netfilter/iptables firewall with
> Linux
> > 2.4.18 + iptables 1.2.5 patch-o-matic (portions of
> > it).
> > 
> > I've been having problems with memory consumption.
> > Basically, as time goes by, 4kb chunks of memory
> are
> > consumed every few seconds.
> [...]
> > Here is some info:
> > 1108 root@firewall:/#cat /proc/meminfo 
> >         total:    used:    free:  shared: buffers:
> 
> > cached:
> > Mem:  261722112 74821632 186900480        0  
> 716800
> > 35336192
> 
> This is obviously from when you don't have the
> problem.
> 
> Best approach is, I hope, to run
> 
>       watch 'sort -nr -k 2 /proc/slabinfo'
> 
> Do you see any of the slabs continually growing?
> 
> You have so many modules loaded, that I won't hazard
> a guess to where
> the memory is going to.
> 
> best regards
>   Patrick



__________________________________________________
Do You Yahoo!?
Yahoo! Sports - live college hoops coverage
http://sports.yahoo.com/

Reply via email to