From: Mulyadi S. <a_m...@te...> - 2004-04-27 09:47:26
|
Hello Luke :-) > Actually, the problem isn't specific to 2.4.22. I've been using oM > about a year now and have always observed "sudden death." The Sorry, i missed that fact :-) > monitoring daemon checks in on the RAID only every few minutes, and > turning off RAID monitoring is a very bad idea. Besides, that can't be > the issue because all the other nodes without RAID controllers crash as > well. OK, in that case, let's throw out the possibility of controllers' weirdness > Given the consistency across kernel versions I think an e1000 problem is > the most plausible explanation that we've come up with. I will try to > get the stock intel driver in there this weekend. That is the last possibility. I hope you can try Intel drivers ASAP :-) > I hope we learn > something, because all my users are irate at all the crashing, and I > probably only have a few weeks left before I'm forced to move to another > load balancing system. For me, i learn much from decoding your "oops". It is not an easy one, but certainly not the most difficult. Everything is leading to ethernet device driver and maybe the RAID controller or magicdev just suffered the side effects About choosing load balancing systems, it is 100% up to you to pick the most suitable :-) Experiment and do stress testing, then you will find one. Just because oM is so far top ranked in open source SSI, that doesn't mean it will be perfectly suited for any case.....:-) I hope my explanation can help you, Luke........ regards Mulyadi |