Re: A per-instance shared queue among forked child processes?

I can't easily use normal (permanent) tables because multiple instances of the script will step all over each other.

Why not just use a permanent table that is associated with each child's process id?

my $pid = fork();

if (not defined $pid) {
    warn "resources not avilable.\n";

} elsif ($pid == 0) {
    print "Child: My PID is $$\n";
    exit(0);
}

print "Parent: I talk to child by adding to QUEUE associated with $pid
+\n";
[download]

Comment on Re: A per-instance shared queue among forked child processes? Download Code

Replies are listed 'Best First'.
Re^2: A per-instance shared queue among forked child processes? by EvanK (Chaplain) on Apr 04, 2011 at 19:57 UTC
That's a great idea that I can't believe I didn't think of. I can create the table in the parent: `$queue_table = 'queue_'.$$; $dbh->do('CREATE TABLE '.$queue_table.' (payload TEXT NOT NULL)');` [download] And then access it from the children after the fork, via `$queue_table`.	[reply] [d/l] [select]
Re^3: A per-instance shared queue among forked child processes? by wind (Priest) on Apr 04, 2011 at 20:27 UTC
I personally would not create a new table for each instance, but would instead have a static table that contains the key PARENTID or CHILDID depending on if you wanted the queues to be associated to the parent or child. `my $sth = $dbh->prepare(q{INSERT INTO TABLE QUEUE SET PARENTID=?, PAYL +OAD=?}); $sth->execute($parentid, $payload) or die $dbh->errstr;` [download] Before each new instance of a parent is run, it should simply scan the table to see if there are any old records left over from a previous process with the same pid.	[reply] [d/l]