[ale] NFS mount problems

[Ryan Fish]

>>>Do the NFS timeouts occur before or after you run mount manually?

The timeout just happens over time throughout the day.  I only tried
mounting them manually last night to see if it would clear things up.  It,
of course, did not.

>>>>What does the mount (and `df -h') output look like after you've done
this manual mount?  Is there an automounter involved somewhere by any
chance?

Before the manual mount "df -h" just hangs when trying to list the mount
points.  After it shows both of them twice.

>>>>Without that further info, it sounds like you have remounted NFS
mounts on top of already mounted NFS mount points.  Which is bad, and
could account for the system load going through the roof.

The load was already going through the roof before I even tried to manually
mount anything.  It just grew until it got high enough to take down the
application that talks to the DB around 2:49A today (the same time it did
this the day before).

>>>>Something more esoteric that could account for the same problem would
be a symlink on the NFS drive that (either directly or indirectly)
points to itself.

The NAS is a W2K3 Std Ed. box that is used to collect backup files from all
servers via bash scripts.  In the case of the Oracle server it is used by
export and RMAN throughout the day to collect both backups of the DB as well
as the arch logs written throughout the day.

>>>Have you analyzed /proc on the oracle server or the NFS traffic
between the two systems to see what files are being read/written over
the connection?

What should I look at/for in /proc?  I have not looked at anything in there
as of yet but honestly have no idea what to look for.

How should I analyze the traffic between the two boxes to see the R/W info?


Thank you.
-Ryan