[Beowulf] Raid disk read performance issue in Linux 2.6, anyone seen this?
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Craig Tierney ctierney at HPTI.comTue Dec 7 11:25:44 PST 2004
- Previous message: [Beowulf] Raid disk read performance issue in Linux 2.6, anyone seen this?
- Next message: [Beowulf] [Lustre-announce] Lustre 1.2.8 is now available (fwd from phil@clusterfs.com)
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Mon, 2004-12-06 at 23:44, Chris Samuel wrote: > On Sun, 5 Dec 2004 04:42 am, Craig Tierney wrote: > > > For the 2.4 kernels, I needed to tweak /proc/sys/vm/max_readahead > > to improve performance. > > Out of interest, what was your tweak ? > > cheers! > Chris The max_readahead is too small for high performance RAID hardware. I typically set: echo 511 > /proc/sys/vm/max_readahead But it depends. If I am using LVM to stripe across volumes I have seen better performance with 1023. I would test out different values, but I have never seen where the default value of 31 was adequate. I also changed /proc/sys/vm/min_readahead, but that didn't seem to make a difference on the large streaming IO opterations. Craig
- Previous message: [Beowulf] Raid disk read performance issue in Linux 2.6, anyone seen this?
- Next message: [Beowulf] [Lustre-announce] Lustre 1.2.8 is now available (fwd from phil@clusterfs.com)
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
