Postgrey crashed, Database lost ? - Benjamin Sonntag Blog

All the versions of this article: [English] [français]

Le 25/1/2006, 00:00
Postgrey crashed, Database lost ?

This morning, the web site was down, certainly due to a ram issue. However, it is now back online.

After the reboot, postgrey, an anti-spam daemon using an automagically filled whitelist database refused to start with the following error in syslog:

ERROR: can’t create DB environment: No such file or directory

I found that the greylisting database in /var/lib/postgrey was corrupted. When I strace postgrey start script with strace -f /etc/init.d/postgrey start I saw misc error messages such as "__db.001: no such file or directory".

After searching the Internet, I found a guy who had the same error and simply destroyed its greylisting berkeley database (sniff, my db ? dead ?). But I didn’t want to loose mine, so I do the following :

# check that db4.3 tools are installed 

# (use your version and package manager for this step)

apt-get install db4.3-util

# dump the current databases : 

cd /var/lib/postgrey

db4.3_dump -p postgrey.db >/var/tmp/postgrey.db

db4.3_dump -p postgrey_clients.db >/var/tmp/postgrey_clients.db

# Kill everybody

killall -15 postgrey 

# (this killall said logically "no process killed")

# destroy the pid file and all the database files 

# to start with fresh entries

rm /var/run/postgrey.pid

rm /var/lib/postgrey/* -f

# start postgrey : 

/etc/init.d/postgrey start

# here postgrey is up and running, (with an emtpy db of course)

# So we stop it : 

/etc/init.d/postgrey stop

# and restore the db from the dump : 

db4.3_load -f /var/tmp/postgrey_clients.db postgrey_clients.db 

db4.3_load -f /var/tmp/postgrey.db postgrey.db 

# that's all, just restart postgrey : 

/etc/init.d/postgrey start