Hi.
So strictly speaking this isn't a ganglia issue per se. But
ganglia caused us to notice it.
We ran ganglia on a bunch of file servers and two batch clusters
one of which has 204 nodes at the moment. Apparently the total number
of nodes put the number over the arp cache limit of our linux server
network and qwe started seeing problems related to arp cache
time-outs.
Now, it occurs to me that there are cluster out there bigger than
ours with thousands or a least many hundreds more nodes than we have.
So the questions are
o Do they have arp cache problems?
o If so what do they do?
This seems like a pretty annoying problem that is very common in
unix/linux systems. There are solutions but I'm more interested in
what others do that works.
Help apprecated. Thank you.
-------------------------------------------------------------------------------
Chris Johnson |Internet: [EMAIL PROTECTED]
Systems Administrator |Web: http://www.nmr.mgh.harvard.edu/~johnson
NMR Center |Voice: 617.726.0949
Mass. General Hospital |FAX: 617.726.7422
149 (2301) 13th Street |Life n, something from which nobody escapes
Charlestown, MA., 02129 USA |alive. Me
-------------------------------------------------------------------------------
