From: CSBVAX::CSBVAX::MRGATE::"SMTP::RESEARCH.ATT.COM::GREGG%IHLPB%ATT" 22-DEC-1988 06:57 To: MRGATE::"ARISIA::EVERHART" Subj: Re: CTLSMB seems to work pretty well Received: from YMIR.BITNET by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with BSMTP id 9261; Mon, 19 Dec 88 14:37:29 EDT Received: from CUNYVM.BITNET by YMIR.BITNET; Mon, 19 Dec 88 07:52 PST Received: from CUNYVM by CUNYVM.BITNET (Mailer R2.01) with BSMTP id 7711; Mon, 19 Dec 88 10:24:30 EDT Received: from research.att.com by CUNYVM.CUNY.EDU (IBM VM SMTP R1.1) with TCP; Mon, 19 Dec 88 10:24:08 EDT Received: by att.ATT.COM (smail2.6 - att-ih) id AA06890; 18 Dec 88 20:37:21 CST (Sun) Date: Sun, 18 Dec 88 20:21:33 -0600 Message-id: <18445.598501293@ihlpb.att.com> From: Gregg Wonderly Subject: Re: CTLSMB seems to work pretty well To: "John L. Jamison x8545" Cc: ipmdf@YMIR.BITNET X-VMS-To: "John L. Jamison x8545" Resent-date: Mon, 19 Dec 88 07:53 PST Resent-message-id: <0091D8EDE5CFE1602040564E@YMIR.BITNET> Resent-to: INFO-PMDF-LIST2@YMIR.BITNET In-reply-to: Your message of Sun, 18 Dec 88 02:33:00 -0500. <0091D7DDA52D5180204000A5@YMIR.BITNET> > I just got Gregg Wonderly's CTLSMB stuff to work, and I'm really > glad to have made the effort- seems like the mail jobs just whiz > through now. It is really phenominal how much faster isn't it? > Error in [CTLSMB.SRC]MAKE.COM Must be a reference to DO.COM. This assignment should fix the problem > $ ASSIGN PMDF_ROOT:[CTLSMB.EXE] PUB > [CTLSMB.EXE]CONFIG.DAT > I didn't touch it, except to delete the IBM$PRINT stuff. I > couldn't make heads or tails of it... The CONFIG.DAT file shipped was mine on NEMO. The IBM$PRINT queue provides RJE access to the Comp Center's 3081K via an AT&T 3B15 with an RJE line. Basically, this queue is tied to a command procedure which mails the submitted file to the UN*X machine. The account there recognizes its incomming mail as a job stream for the IBM. It inserts a JCL comment that identifies the job, and submits it. When the job comes back, the submitters e-mail address is extracted from the comment ID and the output is returned properly. A bit drawn out, but totally effective... No more IBM editing and other such hassles (data set expiration). > [CTLSMB.EXE]CTLSMB_STARTUP.COM > Make sure you execute this command procedure BEFORE you try to > start the queues- seems like the symbiont needs the logical > CTLSMB$ROOT (which is defined in CTLSMB_STARTUP) before it > can do anything. This is why my previous attempts to use > CTLSMB failed: the jobs bomb out without much explanation when > CTLSMB$ROOT is undefined. I needs CTLSMB$ROOT to get to CONFIG.DAT... I thought this was self evident, but obviously not, sorry for the confusion... It has been some time since I set all of this up (the comment on batch_monitor.com says its been over a year since I first wrote it). I don't really know what state things are in in the PMDF distribution, the CTLSMB directory is not on line on NEMO, and that is the only VMS VAX I have access to (dialup at that). The CTLSMB docs where never really proofed and gone over very well, so I appologize again for problems there. > BATCHMONITOR.COM > I just added the line "@CTLSMB$ROOT:[EXE]BATCHMONITOR" inside the > SET NOON lines in POST, MASTER and RETURN (added SET NOON to RETURN). BATCHMONITOR.COM is supposed to execute these procedures for you. It should be the file= value in the config.dat file as in pmdf$queue file=pmdf_root:[exe]batch_monitor.com, output=pmdf_root:[log]localq_start.log, error=pmdf_root:[log]localq_start.log Which will allow batch_monitor.com to look at the submitted .COM files name and redirect the output and other such things to jive with the normal operations of a VMS batch queue. The initial output and error files will leave information about the execution of batch_monitor.com. Below is the current batch_monitor.com file which should be used for PMDF 3.0 and later. Ned and I rearranged the POST.COM, RETURN.COM and MASTER.COM files so that they understand (when necessary) that they are running under CTLSMB. $! $! Command procedure to set in front of POST, MASTER and RETURN. It $! defines the proper logfile redirection since Server queues do not $! support the notion of a logfile. Also, the ON THEN GOTO other $! statements allow any procedure to be envoked within the command file. $! $! Written by Gregg Wonderly 29-OCT-1987 $! $! $ fn=f$PARSE(SMB$FILESPEC,,,"NAME") $ on error then goto other $ on severe_error then goto other $ on warning then goto other $ goto process_'fn' $ process_: $ EXIT $ ! $ process_MASTER: $! $ define sys$output pmdf$root:[log]'p1'_master.log $ define sys$error pmdf$root:[log]'p1'_master.log $ @'SMB$FILESPEC' "''p1'" "''p2'" "''p3'" "''p4'" "''p5'" $ deassign sys$error $ deassign sys$output $ EXIT $! $ process_POST: $! $ define sys$output pmdf$root:[log]post.log $ define sys$error pmdf$root:[log]post.log $ @'SMB$FILESPEC' "''p1'" "''p2'" "''p3'" $ deassign sys$error $ deassign sys$output $ EXIT $! $ process_RETURN: $ define sys$output pmdf$root:[log]return.log $ define sys$error pmdf$root:[log]return.log $ @'SMB$FILESPEC' "''p1'" "''p2'" "''p3'" $ deassign sys$error $ deassign sys$output $ EXIT $! $ other: $ on error then stop $ on severe_error then stop $ on warning then stop $ @'SMB$FILESPEC' "''p1'" "''p2'" "''p3'" "''p4'" "''p5'" "''p6'" $ EXIT > PMDF startup > I added the line: > $ @CTLSMB$ROOT:[EXE]CTLSMB_STARTUP to PMDF_ROOT:[EXE]PMDF_STARTUP That is the correct way to do it... > Periodic job startup > Be sure to make PMDF_SUBMIT_JOBS use the logical name for the queue > (MAIL$BATCH) instead of a hard-coded batch-queue name. This way, you > don't have to change these files if you switch back and forth between > normal BATCH and CTLSMB (I guess I'm just skeptical at this point- I > don't know how its going to work!) just by redefining MAIL$BATCH. Please post any information about problems, and/or send them to me. CTLSMB has been running on NEMO since at least the date shown on batch_monitor.com. There were initially some problems with certain aspects of dealing with the job controller, and asynchronous processing in FORTRAN, but I believe that they are all ironed out now. Again, please post or forward comments to me. CTLSMB is a great way to move a very large number of special purpose batch jobs through the queues with a minimal amount of overhead. ----- Gregg Wonderly DOMAIN: gregg@ihlpb.att.com AT&T Bell Laboratories UUCP: att!ihlpb!gregg