Philip Prindeville a écrit :
I'm running 2.6.20.21 as part of AstLinux 0.50 on a Soekris net5501
(it's a lower power Geode LX 433), and I occasionally have the box go
catatonic and have to be rebooted. Before that happens, dmesg fills up
with:
Feb 19 00:46:22 pbx user.warn kernel: printk: 12 messages suppressed.
Feb 19 00:46:22 pbx user.warn kernel: dst cache overflow
Feb 19 00:46:28 pbx user.warn kernel: printk: 18 messages suppressed.
Feb 19 00:46:28 pbx user.warn kernel: dst cache overflow
Feb 19 00:46:32 pbx user.warn kernel: printk: 19 messages suppressed.
Feb 19 00:46:32 pbx user.warn kernel: dst cache overflow
Feb 19 00:46:38 pbx user.warn kernel: printk: 16 messages suppressed.
Feb 19 00:46:38 pbx user.warn kernel: dst cache overflow
...
What I've got loaded when this happens is:
# lsmod
Module Size Used by
nf_nat_sip 4160 0 nf_conntrack_sip 9236 1 nf_nat_sip
binfmt_misc 10440 1 xt_physdev 2384 0
act_police 5860 1 cls_u32 7684 1
sch_ingress 3168 1 sch_sfq 5344 4
sch_htb 16160 1 xt_length 1696 0
xt_CLASSIFY 1600 0 ipt_tos 1408 0
xt_tcpudp 2880 125 xt_MARK 2048 0
ipt_TOS 1984 34 ipt_MASQUERADE 3648 1
nf_nat_ftp 3136 0 ipt_TCPMSS 3616 3
ipt_LOG 5728 43 ipt_REJECT 4128 0
iptable_nat 7076 1 nf_nat 17676 4
nf_nat_sip,ipt_MASQUERADE,nf_nat_ftp,iptable_nat
iptable_mangle 2528 1 iptable_filter 2688 1
xt_multiport 2816 0 xt_state 2208 14
xt_limit 2400 47 xt_conntrack 2496 0
nf_conntrack_ftp 9088 1 nf_nat_ftp
nf_conntrack_ipv4 16844 16 iptable_nat
nf_conntrack 57496 10
nf_nat_sip,nf_conntrack_sip,ipt_MASQUERADE,nf_nat_ftp,iptable_nat,nf_nat,xt_state,xt_conntrack,nf_conntrack_ftp,nf_conntrack_ipv4
nfnetlink 6328 3 nf_nat,nf_conntrack_ipv4,nf_conntrack
ip_tables 11400 3 iptable_nat,iptable_mangle,iptable_filter
x_tables 13988 17
xt_physdev,xt_length,xt_CLASSIFY,ipt_tos,xt_tcpudp,xt_MARK,ipt_TOS,ipt_MASQUERADE,ipt_TCPMSS,ipt_LOG,ipt_REJECT,iptable_nat,xt_multiport,xt_state,xt_limit,xt_conntrack,ip_tables
pc87360 18288 0 hwmon_vid 2464 1 pc87360
hwmon 3060 1 pc87360
i2c_isa 4928 1 pc87360
i2c_core 20464 2 pc87360,i2c_isa
bridge 48280 0 llc 6996 1 bridge
ztdummy 4520 0 rtc 12336 1 ztdummy
wanec 295224 0 wanpipe_lip 116444 0
af_wanpipe 29184 0 wanpipe 873504 0
wanpipe_syncppp 24724 1 wanpipe
wanrouter 35752 5
wanec,wanpipe_lip,af_wanpipe,wanpipe,wanpipe_syncppp
zaptel 192676 4 ztdummy,wanpipe
hdlc 6180 1 zaptel
ppp_generic 25428 1 zaptel
slhc 6112 1 ppp_generic
sdladrv 55840 2 wanpipe,wanrouter
geode_rng 2144 0 geode_aes 5636 0
blkcipher 5344 1 geode_aes
via_rhine 22184 0 mii 5024 1 via_rhine
cs5535_gpio 4596 0
I'll try to get the contents of /proc/meminfo next time before I reboot
(via the console).
What's going on? I googled for this message and got a lot of hits, but
nothing that seems specific to this issue.
What do I need to capture to get my smoking gun?
Your IP route cache is filled with too many entries.
You can tune it so that less entries are stored in it.
Please post :
grep . /proc/sys/net/ipv4/route/*
rtstat -c10 -i1
Many changes were done these last months on net/ipv4/route.c, so you could try
a recent kernel (2.6.24)
--
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
