Last week I cloned a Linux Squid proxy/cache server ("dd" is your friend) for deployment out at one of our magnet schools that is currently starved for WAN bandwidth. After making the appropriate changes to give the clone its own identity, we hauled it out to the school and set it up. It ran fine for a couple of days. The cache was building, and the ratio of cache hits to cache misses was improving by the hour. Then, on Friday it crashed.
The server was unresponsive when I got to it, so my only real recourse was to power-cycle it. It ran through the weekend, and then crashed Monday morning. It went off into la la land yesterday afternoon (still responding to pings, but otherwise dead). One of my crew went out to the school this morning and found it with a kernel panic message on the screen. I had him tote the server back to our shop for some love and attention.
The first two lockups have generated this single syslog entry: "Dec 11 10:48:11 pm-lnx-squid01 kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000". The last one gave a similar kernel panic message on the console.
Anybody have any clues?
