Data isn't being collected for several metric groups and the [hostname]-bgscollect-noInstance.log, [hostname]-bgsagent_6767.log, and the UCM Status Reports indicate that groups are in ACCOUNT_UNAUTHORIZED state and not being collected.
Further down in the log file, there are authorization errors for every metric group:
NOTE: This document was originally published as Solution SLN000000144354.
The most common cause of the problem is that the Capacity Agent bgscollect binary is not owned by root, setuid root.
If the permissions on the file are not '-r-sr-sr-x' or the owner of the file is not 'root' then the best way to resolve this problem is to re-run the [Installation Directory]/b1configVVVVV.sh script (where VVVVV is the product version, such as 11500) as root. This will correct the permissions of the bgscollect binary and correct any other permissions problem that might exist with the key binaries in the Perform installation.
# cd /[Installation Directory]
# ./b1configVVVV.sh (Where VVVVV is the Agent version (i.e. b1config11500.sh)
Alternately, if you don't want to re-run the b1configVVVVV.sh script or the problem still persists, other possible causes and solutions are discussed below.
> cd $BEST1_HOME/bgs/bin
> ls -la bgscollect
-r-sr-sr-x 1 root root 2367640 Apr 22 12:32 bgscollect
If bgscollect is not owned by root and the permissions are not set properly, run the following commands:
> chown root bgscollect
> chgrp root bgscollect
> chmod 6555 bgscollect
You can also see the permissions of the bgscollect binary in the $BEST1_HOME/bgs/log/[hostname]-bgscollect-noInstance.log file:
Tue Feb 9 14:43:47 2010 bgscollect (5755) Installation info for ../../bin/bgscollect: 4092 -r-sr-sr-x root root 2094584 Jun 12 10:11:43 2008 bgscollect
If the permissions and ownership of $BEST1_HOME/bgs/bin/bgscollect are set properly (per the instructions in Step 1) and ACCOUNT_UNAUTHORIZED errors are still being generated in the [nodename]-bgscollect-noInstance.log file, check /etc/mnttab on the node for the filesystem where the Perform product is installed. The /etc/mnttab file will have entries similar to the following:
/proc /proc proc rw,suid,dev=2a00000 1049401868
If the filesystem where Perform is installed has the "nosuid" option set the collector will fail to gather data. Change the "nosuid" to "suid" for the Perform filesystem, and retry the data collection.
For example, here is the /etc/mnttab entry for a Perform file system where the nosuid flag has been specified:
/dev/vx/dsk/grpspc_share_dg/perf-vol /opt/perform vxfs rw,nosuid,delaylog,largefiles,cluster,ioerror=mdisable,crw,dev=4586593 1260730664
Sometimes the permissions on bgscollect have been corrected on the file system, but the running bgscollect process was started before the bgscollect binary was made setuid root.
To address that issue stop and restart the Perform Agent.
To stop the bgsagent run the command:
> /usr/adm/best1_default/bgs/scripts/best1agent_stop -b /usr/adm/best1_default -a
Wait about 10 seconds, and then check 'ps' output to ensure the bgsagent
and bgscollect processes are not running.
> ps -ef | grep bgs
To restart the Perform Agent:
> /usr/adm/best1_default/bgs/scripts/best1collect -B /usr/adm/best1_default -q
Additional Log Messages
The following message will also be generated to the $BEST1_HOME/bgs/log/[hostname]-bgsagent_6767.log: