From: "Will Honea" Received: from mxout2.mailhop.org ([63.208.196.166] verified) by 2rosenthals.com (CommuniGate Pro SMTP 5.0.9) with ESMTP id 287823 for os2-wireless_users@2rosenthals.com; Sun, 13 Aug 2006 23:24:40 -0400 Received: from mxin2.mailhop.org ([63.208.196.176]) by mxout2.mailhop.org with esmtp (Exim 4.51) id 1GCT48-000Itu-83 for os2-wireless_users@2rosenthals.com; Sun, 13 Aug 2006 23:24:37 -0400 Received: from smtp106.biz.mail.mud.yahoo.com ([68.142.200.254]) by mxin2.mailhop.org with smtp (Exim 4.51) id 1GCT48-0006E9-24 for os2-wireless_users@2rosenthals.com; Sun, 13 Aug 2006 23:24:36 -0400 Received: (qmail 10287 invoked from network); 14 Aug 2006 03:24:30 -0000 Received: from unknown (HELO basement) (whonea@whonea.net@71.34.131.238 with login) by smtp106.biz.mail.mud.yahoo.com with SMTP; 14 Aug 2006 03:24:29 -0000 Date: Sun, 13 Aug 2006 21:24:29 -0600 To: "OS/2 Wireless Users Mailing List" Reply-To: "Will Honea" In-Reply-To: References: Subject: Re: [OS2Wireless]Re: Changed -> SOCKET.SYS crash - REXX Issues MIME-Version: 1.0 X-Mailer: The Polarbar Mailer; version=1.25d; build=2006 X-Mailer-Platform: OS/2; architecture=x86; version=20.45 X-Mailer-Java-VM: IBM Corporation; version=J2RE 1.3.1 IBM build co131-20030618 (JIT enabled: jitc); compiler=jitc X-Mail-Handler: MailHop by DynDNS X-Spam-Score: -2.0 (--) Message-ID: ** Reply to message from "Rick R." on Sun, 13 Aug 2006 19:40:43 -0700 (PDT) > ** Reply to message from "Will Honea" > > >In case it got overlooked, if there is any substantial REXX activity you can > easily get the SOCKET.SYS message - REXX has a bad habit of not closing sockets > until the top level program is closed. > > Aha, THANKS FOR THAT INFO (first piece of good info about this one). > I didn't know that. I do have several REXX monitoring programs running in the > background, to check for system intrusions. monitor file changes and online > encryption. > I also use REXX cron and HW monitoring tools. > All those REXX scripts are very short, but they do run repeatedly. So do you > have any details about this REXX & TCP/IP socket issue? > Very eagerly awaiting any details you got! I'm woefully short on details, but the gist of the problem is that child (REXX) programs don't release sockets they open so that if one script calls a second strip etc. etc. non of the sockets used by the children are released until the parent closes. I can't tell you if this is common to both classic and object REXX or just to one of them, but it might be a place to start. Bob Eager (rde@tavi.co.uk) cna probably give you all the details. The only time I saw it was when I had a script that was tracking events and running all the time. It called sub-programs and would eventually crash. Wound up having to kill it with a cron job then restart it every hour or so until I finally re-wrote the monitor and children in C. It seems that exiting the parent cleans up very nicely, but not the sub-programs. -- Will Honea