[Orca-users] Orca problems across NFS

[pandanus at lineone.net]

I have been happily using Orca on my local server for months.
I have now altered the configuration slightly so that all generated
RRD files and html files will be stored on an NFS share.  It still
works but instead of taking under a minute to generate all of my
reports, it takes almost an hour. At first I thought the process had
hung.

Does anyone know if orca has problems when the output goes to an NFS
share. My NFS link is over a 1GB link so speed is not an issue.

I ran a 'ps -ef | grep orca' over a period of 40 minutes and saw the
following output:

10 perl -w /usr/local/bin/orca -v /usr/local/lib/orcallator.cfg
20 RICHP.se orcallator.se
30 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date1>.gz
40 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date2>.gz
50 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date3>.gz
60 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date4>.gz
70 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date5>.gz
80 gunzip -c /home/iscdev/orca/orca_data/util01/percol.<date6>.gz
..
..

I did a truss on the gunzip processes to find out why they were
hanging and the output I got was:

80:  write(1, "  t i m e s t a m p   l".., 32768) (sleeping)

These processes eventually woke up after about 40 minutes and then
the html output was generated.

Any ideas?