Error DB Logs Size

[David Teh]

Hi folks,

Old topic 😛

Running 5.8.7 on Solaris.

Was alerted that the $HCIROOT directory was taking too much space and found one of the site’s error DB files to be super hugh.

Apparently, one of the outbound threads was mistakenly configured to have an “ack sending” TCL. That resulted in lots of 101 errors. Executing a hcidbdump delete would take hours.

1. Would it be possible to reduce the error DB file sizes without taking down the site? I know it’s a”no” but still wanted to check in case….
2. Initially, not knowing the high number of error message we had, we ran hcidbdump -e -D. Unfortuanately, my session to the server broke. After doing the lmclear stuff, I reran the command. Surprised to see that it started from the earlier messages again?!?! Weren’t these messages already deleted? Strange….

TIA!

 

 

Topic

[Charlie Bursell]

If me, I would shut down input threads and make sure the recovery DB clears then reinit the DB.  That way you know all is good.  I know it is a pain but IMHO the best way.

[David Teh]

Thanks Charlie!

[Rob Abbott]

you can also re-init the error DB only with the following command, this will leave the recovery database intact.

 

hcidbinit -e -C

 

Make sure all engine processes and the lockmanager in the site are stopped before running this command!

Rob Abbott
Cloverleaf Emeritus

[David Teh]

Thanks Rob!

[Author]

Replies