Jeff Templon wrote:
> There are a lot of bdii-fwd processes hanging around ... i am not sure
Those processes correspond to clients that do not go away after doing
their queries; they will be cleaned up after a few minutes, when their
slapd gets restarted (that is one of the nice features of the BDII).
Something else, please change this bit in /etc/init.d/bdii:
-----------------------------------------------------------
export LANG=C
su - $BDII_USER -c "
${BDII_DIR}/sbin/bdii-update ${BDII_CONF} &
echo \$! > $PID_FILE
"
-----------------------------------------------------------
to this:
-----------------------------------------------------------
su - $BDII_USER -c "
export LANG=C
${BDII_DIR}/sbin/bdii-update ${BDII_CONF} &
echo \$! > $PID_FILE
"
-----------------------------------------------------------
to avoid having something like en_US.UTF-8 used as LANG,
which would slow down the bdii-update process by a factor 10.
Next, I recommend you put the BDII on a separate node, or is the
BDII on the RB only used by the RB itself?
> whether it is significant, but ps is reporting a lot of white space in
> the command:
>
>> lcgbdii 20836 0.0 0.1 7980 4440 pts/1 S 13:00 0:00
>> bdii-fwd [192.16.186.252:37581 <--
>> 127.0.0.1:2172]
>> \n
That is a bad but harmless mismatch between perl trying to set argv[0]
to something interesting and ps displaying the result. I do not see
this on SL3.
Also the contents of /opt/bdii/var/bdii.log and bdii-fwd.log look normal.
|