[Orca-users] Orcallator process dies

[Perry Graves]

Hi,

We have just recently installed 4 Sun V880 servers (2 are attached to
3510 array) and installed Orca to help watch performance and trends.  I
had some initial problems with compiling, but looking through all the
email threads helped solve this.  I'm very happy with the collection and
graphs.

The problem I'm experiencing is orcallator will die on one of servers. 
I've attempted to keep versions and any changes the same across the
servers, other than orcallator.se (max_columns and max_rawdisks).  I've
had to increase these values to 8192 on the problem server the others
are set to 4096, so I don't get a segmentation error when starting.  The
apparent reason it dies is max_columns needs to be increased, even more.
 This server has very little activity.

What I have are V880's running Solaris 9, each server has 6 locally
attached disks; (problem server) is a 4way w/32Gb (attaches to 3510),
others are 1- 8way w/64Gb (attaches to 3510), and 2 - 6way w/48Gb.  I've
installed Orca 0.27b3 (orcallator.se 1.37 and orcallator.cfg 1.36),
SEtools 3.4, Active Perl 5.6.1 and gcc 3.3.2.  And added note mpxio is
enabled on the servers,  all 4 servers will be attached to the 3510 at
some point. 

I'm not fully understanding what the max_columns or max_rawdisks
represents.  Maybe someone can clue me in.

Also I have the data files (*.bz2) copied over to my SunBlade for
processing the graphs etc., this is done every 5 minutes.  Once the
files (*.bz2) are processed can they be deleted?

Thanks in advance, any direction will be appreciated.

Perry