public inbox for gentoo-user@lists.gentoo.org
 help / color / mirror / Atom feed
* [gentoo-user] packets with bad cheksum
@ 2004-12-08  9:43 krzaq
  2004-12-08  9:57 ` Jonathan Nichols
  2004-12-08 14:13 ` Jean-Francois Gagnon Laporte
  0 siblings, 2 replies; 8+ messages in thread
From: krzaq @ 2004-12-08  9:43 UTC (permalink / raw
  To: gentoo-user

Hello everybody!

I recently tracked down a problem with my home server.
After running for few hours, all connections start to work faulty (no
connections or slow).
When I run iptraf I see that
approx. 30% of all packets  are classified as BadIP (bad checksum in
header right?).
That concerns all intefraces (WAN, home LAN and even LO!!!!).
Reboot helps but the problem returns shortly.

I've booted form a memtest image and run memtest all day and night.
All tests came out clear.

I have  kernel  2.6.8.1 with some patches (grsec,ip_accounting,imq,...),
but It ran flawlessly for more than a month.

The question is: how the hell can I debug it? Any hints?
Memory is tested and I am quite stucked. I've put down squid and
replaced named with nsd temporarily to
eliminate any CPU & MEM eating daemons -- didn't help.
 
No kernel pacnis, no kprintf's .. nothing in my dmesg ... bummer ... :(

-- 
Regards
Karol Krzak

--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
  2004-12-08  9:43 [gentoo-user] packets with bad cheksum krzaq
@ 2004-12-08  9:57 ` Jonathan Nichols
  2004-12-08 10:08   ` krzaq
  2004-12-08 14:13 ` Jean-Francois Gagnon Laporte
  1 sibling, 1 reply; 8+ messages in thread
From: Jonathan Nichols @ 2004-12-08  9:57 UTC (permalink / raw
  To: gentoo-user


> When I run iptraf I see that
> approx. 30% of all packets  are classified as BadIP (bad checksum in
> header right?).
> That concerns all intefraces (WAN, home LAN and even LO!!!!).
> Reboot helps but the problem returns shortly.
> 

Start looking at the physical layer - replace a cable, see if it helps. 
Sounds like you've already done a lot of software troubleshooting to no 
avail.





--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
  2004-12-08  9:57 ` Jonathan Nichols
@ 2004-12-08 10:08   ` krzaq
  2004-12-08 10:10     ` Jonathan Nichols
  2004-12-08 15:11     ` Billy
  0 siblings, 2 replies; 8+ messages in thread
From: krzaq @ 2004-12-08 10:08 UTC (permalink / raw
  To: gentoo-user

On Wed, 08 Dec 2004 01:57:23 -0800, Jonathan Nichols <jnichols@pbp.net> wrote:
> 
> > When I run iptraf I see that
> > approx. 30% of all packets  are classified as BadIP (bad checksum in
> > header right?).
> > That concerns all intefraces (WAN, home LAN and even LO!!!!).
> > Reboot helps but the problem returns shortly.
> >
> 
> Start looking at the physical layer - replace a cable, see if it helps.
> Sounds like you've already done a lot of software troubleshooting to no
> avail.
Could a faulty cable/switch cause such havock on all interfaces (lo in
particular)?
I am no expert and I have no idea where to start looking. The problem
seems to be
with hardware but not with memory (CPU? , motherboard?).

-- 
Regards
Karol Krzak

--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
  2004-12-08 10:08   ` krzaq
@ 2004-12-08 10:10     ` Jonathan Nichols
  2004-12-08 15:11     ` Billy
  1 sibling, 0 replies; 8+ messages in thread
From: Jonathan Nichols @ 2004-12-08 10:10 UTC (permalink / raw
  To: gentoo-user


> Could a faulty cable/switch cause such havock on all interfaces (lo in
> particular)?
> I am no expert and I have no idea where to start looking. The problem
> seems to be
> with hardware but not with memory (CPU? , motherboard?).
> 

bad ethernet interface, maybe? *shrug* worth a shot


--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
@ 2004-12-08 12:42 brettholcomb
  0 siblings, 0 replies; 8+ messages in thread
From: brettholcomb @ 2004-12-08 12:42 UTC (permalink / raw
  To: gentoo-user

Did you swap out the NIC card and see what happens?

> 
> From: krzaq <krzakers@gmail.com>
> Date: 2004/12/08 Wed AM 09:43:39 GMT
> To: gentoo-user@lists.gentoo.org
> Subject: [gentoo-user] packets with bad cheksum
> 
> Hello everybody!
> 
> I recently tracked down a problem with my home server.
> After running for few hours, all connections start to work faulty (no
snipped some

> The question is: how the hell can I debug it? Any hints?
> Memory is tested and I am quite stucked. I've put down squid and
> replaced named with nsd temporarily to
> eliminate any CPU & MEM eating daemons -- didn't help.
>  
> No kernel pacnis, no kprintf's .. nothing in my dmesg ... bummer ... :(
> 
> -- 
> Regards
> Karol Krzak
> 
> --
> gentoo-user@gentoo.org mailing list
> 
> 


--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
@ 2004-12-08 12:42 brettholcomb
  0 siblings, 0 replies; 8+ messages in thread
From: brettholcomb @ 2004-12-08 12:42 UTC (permalink / raw
  To: gentoo-user

Did you swap out the NIC card and see what happens?

> 
> From: krzaq <krzakers@gmail.com>
> Date: 2004/12/08 Wed AM 09:43:39 GMT
> To: gentoo-user@lists.gentoo.org
> Subject: [gentoo-user] packets with bad cheksum
> 
> Hello everybody!
> 
> I recently tracked down a problem with my home server.
> After running for few hours, all connections start to work faulty (no
snipped some

> The question is: how the hell can I debug it? Any hints?
> Memory is tested and I am quite stucked. I've put down squid and
> replaced named with nsd temporarily to
> eliminate any CPU & MEM eating daemons -- didn't help.
>  
> No kernel pacnis, no kprintf's .. nothing in my dmesg ... bummer ... :(
> 
> -- 
> Regards
> Karol Krzak
> 
> --
> gentoo-user@gentoo.org mailing list
> 
> 


--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
  2004-12-08  9:43 [gentoo-user] packets with bad cheksum krzaq
  2004-12-08  9:57 ` Jonathan Nichols
@ 2004-12-08 14:13 ` Jean-Francois Gagnon Laporte
  1 sibling, 0 replies; 8+ messages in thread
From: Jean-Francois Gagnon Laporte @ 2004-12-08 14:13 UTC (permalink / raw
  To: gentoo-user, krzaq

Hi !

If by any chance you are using the 8139too driver with 2.6.8.1 ...
drop it. This driver is horribly broken and has been flagged as so in
the lkml. It has been fixed in the 2.6.9 series and the problem is not
found in the 2.6.7 series since this comes from a patch that was
merged in the 2.6.8.1

Cheers !

Jean-Francois


On Wed, 8 Dec 2004 10:43:39 +0100, krzaq <krzakers@gmail.com> wrote:
> Hello everybody!
> 
> I recently tracked down a problem with my home server.
> After running for few hours, all connections start to work faulty (no
> connections or slow).
> When I run iptraf I see that
> approx. 30% of all packets  are classified as BadIP (bad checksum in
> header right?).
> That concerns all intefraces (WAN, home LAN and even LO!!!!).
> Reboot helps but the problem returns shortly.
> 
> I've booted form a memtest image and run memtest all day and night.
> All tests came out clear.
> 
> I have  kernel  2.6.8.1 with some patches (grsec,ip_accounting,imq,...),
> but It ran flawlessly for more than a month.
> 
> The question is: how the hell can I debug it? Any hints?
> Memory is tested and I am quite stucked. I've put down squid and
> replaced named with nsd temporarily to
> eliminate any CPU & MEM eating daemons -- didn't help.
> 
> No kernel pacnis, no kprintf's .. nothing in my dmesg ... bummer ... :(
> 
> --
> Regards
> Karol Krzak
> 
> --
> gentoo-user@gentoo.org mailing list
> 
>

--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

* Re: [gentoo-user] packets with bad cheksum
  2004-12-08 10:08   ` krzaq
  2004-12-08 10:10     ` Jonathan Nichols
@ 2004-12-08 15:11     ` Billy
  1 sibling, 0 replies; 8+ messages in thread
From: Billy @ 2004-12-08 15:11 UTC (permalink / raw
  To: gentoo-user

krzaq wrote:

> Could a faulty cable/switch cause such havock on all interfaces (lo in
> particular)?
> I am no expert and I have no idea where to start looking. The problem
> seems to be
> with hardware but not with memory (CPU? , motherboard?).

A friend of mine is a network admin for a rather large university. In 
his office on his wall, he has framed two items.

* an ethernet card
* a blue ethernet cable

Both caused system wide network outages at different times. The ethernet 
card, because it started spewing an invalid MAC address, and overloaded 
all the switches in a boardcast storm. The ethernet cable because of a 
break inside the cable that caused noise on a main interface to one of 
the school's core routers.

Never underestimate the small things. First rule of troubleshooting:

* always try the simple fix first

Change cables, swap ethernet cards

As far as your specific problem goes, I'd try with a vanilla kernel and 
see if your problems persist. Second rule of troubleshooting:

* Always define a control state/configuration

If it works, slowly add your patches till it breaks. If it doesn't make 
a difference, then figure out your last update or change, and 
remove/disable it from the system. Keep doing that until your problem 
goes away.

Third rule of troubleshooting:

* Even unrelated changes can have unforeseen consequences.

There is an urban legend of an auto company Engineer that was sent out 
to a customer's home, because the new car they bought was acting up. The 
owner insisted that the car didn't like vanilla ice cream, because if he 
went to the store and bought any other type of ice cream - the car 
started right up. If he bought vanilla ice cream, the car would not start.

According to the story, the vanilla ice cream was kept in the front of 
the store. If the owner bought vanilla ice cream, it took him less time 
to be in the store, and the engine of the car wasn't able to disipate a 
vapor lock issue. Any other ice cream, and the time was sufficient to 
disipate the problem and have the car start with out problem.

http://www.snopes.com/autos/techno/icecream.asp
http://en.wikipedia.org/wiki/Fuel_pump

--
gentoo-user@gentoo.org mailing list


^ permalink raw reply	[flat|nested] 8+ messages in thread

end of thread, other threads:[~2004-12-08 15:11 UTC | newest]

Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-12-08  9:43 [gentoo-user] packets with bad cheksum krzaq
2004-12-08  9:57 ` Jonathan Nichols
2004-12-08 10:08   ` krzaq
2004-12-08 10:10     ` Jonathan Nichols
2004-12-08 15:11     ` Billy
2004-12-08 14:13 ` Jean-Francois Gagnon Laporte
  -- strict thread matches above, loose matches on Subject: below --
2004-12-08 12:42 brettholcomb
2004-12-08 12:42 brettholcomb

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox