11-12-2013 07:50 AM
This issue is a reproducible scenario where transferring a file between one HPUX node and another using rcp is very slow only for the oracle UNIX account, but fast for other UNIX accounts. Network port speeds/duplex have been reviewed.
We specify a location using rcp so it's not a specific desitination directory.
Thanks in advance!
Solved! Go to Solution.
11-14-2013 06:05 AM
A little more info:
Same rhosts file in use
We traced the network whlie this was happening and only noticed fewer packets per minute. Not a tcp negociation or xfer problem.
From a high level the only differences we can come up with are the group membership, ulimit and home dir.
11-16-2013 02:37 PM
If the same destination is not slow for other users, how about the source?
You mentioned the slowdown appears when running as the oracle user. If the source is on the same physical disk/LUN as the database, and the database is very busy, the disk/LUN might spend most of its time with performing the I/O operations of the database, and only occasionally (in computer timescale) have enough free time to read data for the rcp command.