Which suggests that that particular svn snapshot did handle the upgrade correctly. I wonder if a subsequent upgrade will fix the issue, and if not, how we can solve this. (I have upgraded couchdb many times on this system, and never manually configured logrotate.)
This does seems like it's an upgrade where the old logrotate entry was not overwritten correctly.
Mine (/etc/logrotate .d/couchdb) has:
/var/log/ couchdb/ 0.10.0~ svn818859/ *.log {
delaycompress
weekly
rotate 10
copytruncate
compress
notifempty
missingok
}
Which suggests that that particular svn snapshot did handle the upgrade correctly. I wonder if a subsequent upgrade will fix the issue, and if not, how we can solve this. (I have upgraded couchdb many times on this system, and never manually configured logrotate.)