[ previous ] [ next ] [ threads ]
 
 From:  "Josh Simoneau" <jsimoneau at lmtcs dot com>
 To:  <m0n0wall at lists dot m0n0 dot ch>
 Subject:  RE: [m0n0wall] Beta 1.2b10 - All mbuf clusters exhausted
 Date:  Mon, 19 Sep 2005 11:56:20 -0400
James,
 
Are you noticing that the interface just stops working for a little while after you get this error?
If not, that might be happening soon.
 
How much memory is in the system? What hardware are you running?
 
Generally the system will queue traffic up in a buffer, then once that fills up it'll start filling
up system memory. Once that fills up, you'll get the mbuf error you see. In some cases its the
systems way of saying "I don't have anywhere else to buffer this traffic before I can push it
through the line". At this point the interface will just stop talking, as it's got no place to hold
any more data, and won't talk again until it's free'd up some buffer space. I've seen this happen in
companies where a system or two will have a virus that sends out massive amounts of traffic and the
outbound connection isn't fast enough to move it all in time.
 
The system should be dynamically able to resize the buffer for you and utilize all of the systems
memory if needed. There are ways to force it to use a certain amount of memory, but that really
shouldn't be necessary if you've got the right hardware for the job.
 
Why is it happening on this version and not the other? Its possible that the new version uses a
little more memory, and old version had just enough free to buffer?
 
So, this isn't all that helpful, but at least we can try to figure out why this is happening.
 
Regards,
Josh Simoneau
Engineer, Stripper

________________________________

From: James Sheridan [mailto:stuff at phosphor dash ink dot com]
Sent: Mon 9/19/2005 10:37 AM
To: m0n0wall at lists dot m0n0 dot ch
Subject: Re: [m0n0wall] Beta 1.2b10 - All mbuf clusters exhausted



Anyone?  Thx.

--
James

> Just updated from 1.2b3 to the newest beta and noticed this in my logs
> (many times) during high-volume traffic.
>
> /kernel: All mbuf clusters exhausted, please see tuning(7).
>
> I'm running off a CF card and updated via the GUI update page.
> Everything else looks ok and the device is running so I'm not real
> worried, but obviously something is wrong. Going back to 1.2b3 (where I
> am now) shows no such errors.
>
> Running "df -k" through exec.php on 1.2b3 shows:
>
> $ df -k
> Filesystem 1K-blocks Used Avail Capacity  Mounted on
> /dev/md0c      10879 9669  1210    89%    /
> procfs             4    4     0   100%    /proc
> /dev/ad2a       5919 5553   366    94%    /cf
>
> and on 1.2b10:
>
> $ df -k
> Filesystem 1K-blocks  Used Avail Capacity  Mounted on
> /dev/md0c      10879 10134   745    93%    /
> /dev/ad2a       5919  5613   306    95%    /cf
> procfs             4     4     0   100%    /proc
>
> Am I just at the limits of the card now? Is there a configuration change
> I can make to clear this up?
>
> Thanks.
>
> --
> James
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: m0n0wall dash unsubscribe at lists dot m0n0 dot ch
> For additional commands, e-mail: m0n0wall dash help at lists dot m0n0 dot ch
>



---------------------------------------------------------------------
To unsubscribe, e-mail: m0n0wall dash unsubscribe at lists dot m0n0 dot ch
For additional commands, e-mail: m0n0wall dash help at lists dot m0n0 dot ch