kernel: dst cache overflow

Fedora Fedora at InterCept.Net
Wed Dec 8 19:01:58 UTC 2004


In doing a quick with the 2.4.28 kernel, after flushing the cache with
ip route flush cache, 5mins later we get the infamous "dst cache
overflow" messages. Doing a route -Cn | wc results in the following on
each time the messages started to appear:

 

5mins

4095   29609  315201

 

5mins

4093   29575  315043

 

Our /proc/sys/net/ipv4/route/max_size is set to 4096, so I guess that's
why we get the messages, but eventually the network connections stop and
the firewall needs to be hard rebooted. So I'm not sure of the point in
setting that, unless there is a memory leak somewhere... 

 

Thanks,

Michael Brown

  _____  

From: fedora-list-bounces at redhat.com
[mailto:fedora-list-bounces at redhat.com] On Behalf Of Fedora
Posted At: Wednesday, December 08, 2004 1:32 PM
Posted To: Fedora
Conversation: kernel: dst cache overflow
Subject: kernel: dst cache overflow
  

I have been searching all over the internet for a solution to the common
"kernel: dst cache overflow" problem people have been having... Here's
my synopsis, I am running 14 Linux firewalls, 9 on Fedora core 2, 1 on
Fedora core 1 and 3 on Red Hat 9. All of the firewalls have about 3-4
Copper Gig NICs, on Dell PowerEdge 700 with 512mb memory and a P4 2.8Ghz
CPU. The firewalls have just above bare minimal pkgs installs, all
installed from a kickstart file and all services (if any un-used) have
been turn off and disable in the runlevels. These are only firewalls for
internet and private traffic across T1's and higher.  All fedora core 2
firewalls have and running the latest kernel 2.6.9-1.6_FC2smp, but each
kernel before that also had the errors, so there are no fixes in the
kernel. All Red Hat firewalls are running 2.4.25 kernel and the 1 fedora
core 1 is running 2.4.26, but none of them are experiencing this issue.

 

On the Fedora Core 2 firewalls, I get the below kernel messages after a
certain amount of traffic, what that amount is, I do not know. 

 

kernel: dst cache overflow

kernel: printk: 52 messages suppressed.

kernel: dst cache overflow

kernel: printk: 36 messages suppressed.

kernel: dst cache overflow 

 

After so much traffic or time, the firewall becomes unresponsive on the
network and I may or may not get this new message, and the system needs
a hard reboot.

syslogd: sendto: No buffer space available

kernel: dst cache overflow

last message repeated 5 times

last message repeated 9 times

last message repeated 3 times

syslogd: sendto: No buffer space available

kernel: dst cache overflow

last message repeated 3 times

last message repeated 3 times

 

I have tried just resetting the network ( service network restart), I
also tried just to stop the network for a few hours, I tried the
recommendation of setting the route max_size in
/proc/sys/net/ipv4/route/max_size
(http://seclists.org/lists/vulnwatch/2003/Apr-Jun/0076.html), none
seemed to help. I am now trying a custom 2.4.28 kernel on one of the
firewalls to see if that may help. 

 

Has anyone run into this and fixed it? It seems to be only with Fedora
Core 2, I am curious if Fedora 3 would make a difference, but I am not
ready to leap into that bird nest just yet.

 

 

Thanks,

Michael Brown

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/fedora-list/attachments/20041208/fbf0bfc3/attachment-0001.htm>


More information about the fedora-list mailing list