Recovery of threads outside the engine
A flag in hcidbdump forces all corrupt messages from the recovery to the error database to state 500 for all threads in a site.
hcidbdump -r -U userid
-U userid
permits the user to give a new user ID for accessing
the database with hcidbdump. With this usage, you
can move the corrupt messages to the error database. The owner of the error database
messages is userid.
This command usage scans messages in the recovery database with user userid. If corrupt messages are found during scanning, then hcidbdump automatically recovers with user userid.
When running this command on a corrupt database, these should be observed:
- Back up the database and ensure no engine process is running on the site.
- Because the database is corrupt, it may encounter a raima system error (-900 ~ -966) during the automatic recovery. In this case, hcidbdump panics to prevent additional errors accumulating and eventually corrupting the database again.
- If the error of the corrupt message
is Read msg data chain but mid not exist; err 2,
then run the keybuild rlog command first to
rebuild the key file for the recovery database. After that, run
hcidbdump -r -U userid
again.
Note: If there is a -944 error when
running keybuild, then delete the $DBTAF/rdm.taf file. Then run keybuild rlog again.