the MARK is just syslogd telling you it's still alive. The best (only?) way to know what the problem with your network is is to set up some sort of network monitor, and see what happens around lunch time. I've used ntop for this, but I'm sure there are other. If there's no unusual net activity, then you can start looking at patches, but with such an old kernel, ,perhaps an upgrade to something newer, like 2.2.19 would be a better choice. One idea that springs to mind, if the notwork monitoring shows nothing wrong, is that perhaps there are a lot of dead connections hanging. a network restart would free up the resources so the machine could get back to normal. regards Anders On Thursday 06 September 2001 20.25, Alfredo Cole wrote:
Hi:
I have a server running kernel 2.2.14-5.0smp on a dual Pentium III 550 Mhz, 512 Mb RAM and three 18.2 Gb SCSI RAID5. NIC is a Kingston card. Around 300 clients use this server for mail and samba services. At about 11:00 am every day, the network slows down so much, that users connected through a router can not even ping the machine. Local in-house users can continue to access the server, albeit much slower. Two other servers, a AS/400 and a RS6000 connected to the same network keep going as usual. Executing a network restart solves the problem.
I have investigated on Deja News and it seems other people have reported a similar problem but no concrete solutions. I found an article at http://www.icase.edu/coral/LinuxTCP2.html and they mention a patch for several versions of 2.2.x kernel, including this one. They claim this could solve the problem.
I have looked into /var/log/messages and don't see anything unusual, except for a line:
Aug 27 08:40:45 gponce -- MARK --
which I don't know what it means.
Has anybody had a similar problem, and if so, could you then point the way?
Thank you.