Hello all, This is possibly (even probably) offtopic.
We have a very strange problem here with one of our servers: The box: A dual PIII 866Mhz, Adaptec 7892 160Mb/s scsi card, IBM 20G drive, 3 Intel EEPro100 nics (82557), 512 Mb RAM (133MHz), GigaByte 6VXDC7 rev 1.0 Mainboard. This box is our backup node in a simple cluster using heartbeat and DRBD. It functions as the failover node for our webserver, our proprietry server, and our MySQL DB. Two of the NICS are for heartbeat (to different parts of the cluster). We installed the cluster before we had our UPS delivered. The day the UPS arrived the building the cluster was in had a power surge - 2 hours before we got the UPS. Murphies Law struck! We replaced the power supply and the mainboard in this machine. It had random lockups, randomly lost a nic, all sorts of strange things, but with the new mainboard we have one strange problem left.... The symptoms: 1st of all, it doesnt matter what arrangements of NICS we use. now to the nitty gritty... With 512 Meg of RAM, the box boots, but eventually has a random lockup. Video is gone, nothing works. With 128 Meg of RAM the box boots and runs OK. I'd hate it to be in the cluster with 128Meg of RAM - if it had to do a takeover it would probably grind to a halt . With 256 Meg of RAM it wont boot. It gets past the memory test, and then hangs on the WAIT... bit. This can be with either 256 stick or with 2 128s, in either order. If we take out the SCSI card the box gets through POST with any amount of RAM. Doesnt boot of course lol. But if we put the card in a windoze box with the same RAM and mainboard there is no problem.. Obviously we are thinking that the SCSI card is cactus, but why the behaviour varies according to the amount of RAM in the system has us mystified. Can anyone shed any light as to what might be going on here? Thanking everyone for their time in advance, John P Foster, http://www.golden-orb.com
