While running the "brutal" (-- Alan Cox) Cerberus test suite on a
2.2.17pre10 kernel, we've been seeing pauses of from five to twenty
seconds every two minutes or so. As best I can tell from SysReq, when
the kernel pauses, it's spending time in shrink_mmap(). When pauses
occur, free RAM is down to about 1M; but it's not a true OOM, since
there's always lots of free swap space. (Cerberus scales its memory
usage to avoid filling swap.)
Good News: The pauses are *almost* completely gone after applying
Andrea's patch entitled "VM-global-patch-1". The kernel still pauses,
but only very occasionally, and for no more than two to four seconds;
and even during the pauses, interactive performance is reasonable.
According to Jason Collins, creator of Cerberus, it's at least as good
as any previous 2.2.x when subjected to a Cerberus attack.
Thanks, Andrea!
PS: Anyone who wants to stress the VM, I recommend trying Cerberus
as part of your testing mix. Get it from SourceForge:
<URL:https://sourceforge.net/projects/va-ctcs/>
Testing Details:
Kernel: 2.2.17pre10
Config: SMP, 2G, NFS root
Hardware: Intel Lancewood, two P3s, 512M RAM, DAC960 RAID
Relevant Patches:
Dave/Trond/Neil's NFS merge
Andrea's VM-global-patch-1
Andrea's BIGMEM
Solar Designer's security
Probably Irrelevant Patches:
Rik's fair share scheduler (but disabled at the time)
Andre Hedrick's IDE
Mingo's RAID
VA's E820 memory probe
Backport of 2.4's AGP driver
-- Chip Salzenberg - a.k.a. - <chip@valinux.com> "I wanted to play hopscotch with the impenetrable mystery of existence, but he stepped in a wormhole and had to go in early." // MST3K- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Fri Jul 07 2000 - 23:00:20 EDT