http://pastebin.org/76416
xrx output for 37 nodes. On Thu, Jan 14, 2010 at 1:45 PM, chris kinney <[email protected]> wrote: > tried using xrx with 109 backend nodes and got memory issues... > http://pastebin.org/76404 > > i ran xrx using 38 nodes and it ran the command fine but after it was done > it just halted i had to ctl-z to get out. > > On Thu, Jan 14, 2010 at 10:02 AM, chris kinney <[email protected]>wrote: > >> it works on 2.6.31 with no changes made to the kernel between 2.6.31 and >> 2.6.30. i used 1 backend node running xcpufs and 1 client node running xrx. >> >> >> On Wed, Jan 13, 2010 at 3:07 PM, ron minnich <[email protected]> wrote: >> >>> A little more work shows that if we do this: >>> xrx node /bin/ENOENT >>> >>> Things don't fail. It's when files start to get pulled over that it blows >>> up. >>> >>> It does look like huge bits of the kernel data space are getting so >>> completely trashed that the kernel can't even get through printing the >>> oops info. >>> >>> We're trying 2.6.31 now that was known to work in 32 bit mode on a laptop >>> >>> ron >>> >> >> >
