On Wed, Feb 3, 2016 at 6:39 PM, Greg Sabino Mullane <[email protected]> wrote: >> If anyone sees this, it's because the postgres DBD object went out of >> scope and was disconnected underneath me (the script forks). >> >> The fix was to simply use the "InactiveDestory => 1" as an option at >> the time of dbconnect() as per the doc. Wow, that was painful. > > Glad it got solved! I didn't reply as I couldn't figure out what was > going on, but in retrospect forking is always a good suspect for > mysterious DBI issues (similar to the way SELinux is for > mysterious Linux issues :)
No problem Greg. Pain in the arse though since there were no error messages until I started to use the trace stuff. (I wasted a lot of time trying to turn on postgres logging to convince myself the SQL was never actually hitting the server to begin with!). -aps
