Hi All We've got a Linux Redha

[drucej31]

Hi All

We've got a Linux Redhat 7.3 server, it basically locked up this morning could'nt get access ssh / web etc.. Ended up having to reboot the box manually.. in the /var/adm/messages file we found a number of these error messages..:-

ip_conntrack: table full, dropping packet

and significant volumes of these messages:-

Nov 10 04:05:39 actionaid2 kernel: TCP Dropped IN=eth0 OUT= MAC=00:50:8b:ec:9d:e8:00:04:c1:6d:a9:00:08:00 SRC=65.122.105.2 DST={our server ip} LEN=60 TOS=0x00 PREC=0x00 TTL=51 ID=10694 DF PROTO=TCP SPT=56928 DPT=113 WINDOW=5840 RES=0x00 SYN URGP=0

Does anyone have ideas on what might be causing it.. Having rebooted it has initially solved the immediate problem but concerned it may return..

Thanks in advance.
J.

 

[Cadwalader]

Hello J.

Just a few questions...

1. What else is on the network?
2. What else is going on in your network?
3. What are the hardware details of the server?

I have a RedHat 7.3 box on a network with a Windows 2000Pro box. During a LAN party one night, I used the net send command to *.*.*.255 in a batch file (devilish gigle).
Anyway, after the trash talk, and boasting, I noticed that my poor server didn't like the net send command constantly hitting it. It had some sorta error messege scrolling on the screen. I can't remember what it said, but I had to reboot, and that fixed it. A few weeks later I found out that if I left it running, it may have cleared out. Possibly because the CPU was pegged.

Of course, my box has a abicus for a CPU. That might have has some thing to do with it.... ;-) Hope I was of some help...
--OR--
Thanks for the help...
--Rich

 

[drucej31]

Ok. nothing much really happening on the server it runs Apache / Mysql .. with 3 websites.. Nothing too hefty in fact very light usage, the server is hosted with an ISP in Manchester..

Someone indicated that it might be IP chains related.. but not sure .. we've disabled it just to check whether the problem disappears and attempts to narrow down the problem area..

Also, one of the interesting side effects to this was that the box seemed to want to stream network requests from all the other servers on the same subnet.. very puzzling.. only way of resolving it was to disable the switched port on the hub and reboot the server manually..