Re: Using ProcSignal to get memory context stats from a running backend

Another simpler option would be to open up a new file in the log
directory something like "debug_dump.<pid>.txt" and print whatever you
want there. Then print out a reasonable size log entry saying "Debug
dump output to file 'debug_dump.<pid>.txt'". You could provide a
function that reads such files out of the log directory or just
document how to access them using the pg_read_file().

It's not perfect but it's got most of the advantages of communicating
with the requesting process without the complexities of a DSM segment
and it's a bit more flexible too. Users can have automated monitoring
tools watch for dumps for example. And regular tools can be used to
back up and clean out old files.