not sure i understand what's going wrong (looks like it does find all
the required libraries)...
could you run xcpufs with debug retain-tmp-dir flags and paste the
xcpufs output from the client node?
./xcpufs -D -d -r
and also check the /tmp/xcpu-XXXXXX directory on the client node to see
if it looks ok?
thanks!
On Mon, 2008-10-20 at 17:05 -0600, Abhishek Kulkarni wrote:
> No my bad.
>
> this is something else.. it cannot find the libraries required to
> run /bin/date.
>
> let me look more into this
>
> open("/tmp/xcpu-BBax5M/tls/x86_64/librt.so.1", O_RDONLY)
> = -1 ENOENT (No such file or directory)
> stat("/tmp/xcpu-BBax5M/tls/x86_64", 0x7fff76cfbaf0) = -1
> ENOENT (No such file or directory)
> open("/tmp/xcpu-BBax5M/tls/librt.so.1", O_RDONLY)
> = -1 ENOENT (No such file or directory)
> stat("/tmp/xcpu-BBax5M/tls", 0x7fff76cfbaf0) = -1 ENOENT (No
> such file or directory)
> open("/tmp/xcpu-BBax5M/x86_64/librt.so.1", O_RDONLY)
> = -1 ENOENT (No such file or directory)
> stat("/tmp/xcpu-BBax5M/x86_64", 0x7fff76cfbaf0) = -1 ENOENT (No
> such file or directory)
>
> On Mon, 2008-10-20 at 15:52 -0700, Don W Rudish wrote:
> > We (Chris Kinney and I) are having problems running xcpufs our Perceus
> > cluster. Attached is an strace from the client node. Below are the steps
> > we took:
> >
> > We downloaded xcpu from % svn co svn://xcpu.org/sxcpu/trunk sxcpu
> >
> > Our control node is 10.1.252.10
> > Our client node is 10.1.5.1
> > Both are 64bit RedHat 5 machines
> >
> > ssh to client and started xcpufs with
> > strace -ff -o /tmp/out ./xcpufs
> >
> > from our control node we did:
> > ./xgroupset add 10.1.5.1 root 0 root
> > ./xuserset add 10.1.5.1 root 0 root /id_rsa.pub
> > ./xrx 10.1.5.1 date
>
