Klaus Reimer wrote: > What exactly means "dst cache overflow"? And what is "NET: 85 messages > suppressed"? Maybe I need to increase some logging level to see these > suppressed and maybe important messages?
I have googled a little bit for "dst cache overflow". I found the hint to check this: cat /proc/slabinfo | grep dst I think this is the statistic of the dst cache which may cause the trouble here. I also found proc files to control the maximum number of entries and the gc behaviour. I reduced the maximum number to 4000 and watched the counter. When it reaches 4000 I get some dst cache overflow messages in syslog and the router dies. I then compared the default value of max_size on the router (which is 8192) with the default value which is set on my workstation which has 524288. This is a drastic difference. Maybe this problem is just normal and the maximum number of dst_cache entries is way too low? Maybe it's just a matter of tuning the values of /proc/sys/net/ipv4/route/max_size and /proc/sys/net/ipv4/route/gc_timeout. I set the max_size now to 16384 and reduced the gc_timeout from 300 to 60 and will continue leaching and see if it's better now. -- Bye, K <http://www.ailis.de/~k/> [A735 47EC D87B 1F15 C1E9 53D3 AA03 6173 A723 E391] (Finger [EMAIL PROTECTED] to get public key) _______________________________________________ freewrt-users mailing list freewrt-users@freewrt.org https://www.freewrt.org/lists/listinfo/freewrt-users
