rlnx04 and rmds00 (fwd)

[Dave Zimmerman <zimm@rcf.rhic.bnl.gov>]

---------- Forwarded message ----------
Date: Sat, 03 Oct 1998 12:52:24 -0500
From: "Maguire, Charles F" <charles.f.maguire@vanderbilt.edu>
Reply-To: rhic-rcf-l@bnl.gov
To: RCF List Server <rhic-rcf-l@bnl.gov>
Cc: phenix-off-l <phenix-off-l@bnl.gov>
Subject: rlnx04 and rmds00

Dear RCF staff,

	The rlnx04 platform does not see any directories in the 
/phenix/mds location, or /star/mds.  This should be fixed so that 
rlnx04 can be used to access files on the old MDS whenever the old MDS 
returns to accessible service.

	Since its reboot at about 11:30 AM this morning, the rmds00 
server looks to be pinned at ~99% I/O wait state, such that a 
even a directory listing of an MDS area (e.g. /phenix/mds or /star/mds)
cannot be completed from the rlnx01 machine.  If one logs in directly 
to rmds00, the directory listings works normally.  For all practical 
purposes, files on the MDS are inaccessible in finite amounts of time 
except when looked at directly using rmds00.

	Is the 99% I/O wait state and its apparent effect on the NFS 
due to MDC1 activity or some other present operations on the old MDS?  
If so, when to you think the old MDS might get its I/O wait state out 
of the stratosphere?

Thanks,
Charlie

-----------------------------------------------------------------
Maguire, Charles F
Vanderbilt University
Email: charles.f.maguire@Vanderbilt.Edu