Jim Salter <jim <at> jrssystems.net> writes: > > IANALG (I Am Not A Linux Guy - FreeBSD is my *nix of choice), but could > this possibly be related to the mem:remap Linux kernel bug that was > recently discovered? > > Jim Salter >
It just might have something to do with this. I have come to the conclusion that there is a problem with Linux releaseing cache memory back to processes. Being that I am not a Linux kernel guru, I am about to go through the kernel sources and make environment to see if I can disble, or severely limit, disk cacheing on these boxes. With a 400+ GB database on these systems, the "cache hit" success rate (even with a 4GB cache) would be near nil in any event (1% on a good day with a 4GB cache), making any cache just wasted space. I would much rather get input from someone who has done this before and knows what the end results are...doing it myself is the least favored solution, since this will be a well exposed production resource. This may be of interest to the rsync group, if they start having machines crash during heavy rsync activity they need to know it is not necessarily an rsync problem - rsync just exposed a flaw in the OS. Cheers Kelly -- To unsubscribe or change options: http://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
