Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
Let's Talk About RSCS
I want to be able to submit jobs to JES2 (MVS 3.8j TK5) from VM/370 1.2 CE, but don't want to run MVS as a guest under VM/370.? I get much better performance running a separate Hercules SDL instance for each, and also can use 2 CPUs for MVS.
?
First I made sure RJE is working fine under MVS.? It is, and I can submit jobs and get output from RJE80 running under either Windows or Linux/Wine.
?
Next I configured RSCS (the one distributed with CE).? This connects just fine to MVS.? It took some head scratching, but determined that:
?
START NODEA PARM HASP
?
is what I needed to make the connection from the RSCS end.? I do the following:
?
TAG DEV PUN NODEA
SPOOL PUN RSCS
PUN TEST JCL
?
and my job immediately runs over on the MVS side.? However, the print dataset is not able to make the return trip.? The line "jams" as follows:
?
$HASP094 I/O ERROR ON LINE2??? 671,02,0C00,003D,B400
$HASP094 I/O ERROR ON LINE2??? 671,02,0E00,023D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0C00,003D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0E00,023D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0C00,003D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0E00,023D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0C00,003D,B400 $HASP094 I/O ERROR ON LINE2??? 671,02,0E00,023D,B400 ?
However, if I include a /*ROUTE PRINT RMT1 to send the output to my RJE80 remote, all is well.? This is just a guess, but perhaps there is too much data too fast coming back over the line from JES2 to RSCS and the code cannot handle it so it throw errors.
?
Anyway, I'd like for this to work better, and if I can provide any test data to someone capable and willing to persue the challenge, I'd be happy and greatful.
?
Dan |
On Mon, Oct 7, 2024 at 01:16 PM, Daniel L. Srebnick wrote:
Next I configured RSCS (the one distributed with CE).? This connects just fine to MVS.? It took some head scratching, but determined that:If you'd like to save some of us from losing more hair ;-) would you share your hercules configuration lines for VM and for MVS? ?
I wasn't sure if I needed TCPNJE for both, either, or neither.
?
?... Mark S. |
The scenario posted is without tcpnje. On the MVS side we have: # RJE 0670 2703 dial=IN lport=3780 0671 2703 lport=3781 rhost=localhost rport=3880 dial=no Device 671 is the RSCS connection. On the VM side we have: # RJE 0040 2703 lport=3880 rhost=localhost rport=3781 dial=no On Thu, 2024-10-10 at 11:06 -0700, Mark A. Stevens via groups.io wrote:
|
On Mon, Oct 7, 2024 at 01:16 PM, Daniel L. Srebnick wrote:
However, if I include a /*ROUTE PRINT RMT1 to send the output to my RJE80 remote, all is well.? This is just a guess, but perhaps there is too much data too fast coming back over the line from JES2 to RSCS and the code cannot handle it so it throw errors. IIRC when you submit from a remote station (VM, or RJE) and you want the output back to that station, you need the /*ROUTE as you specified, otherwise it will attempt to print on the "local" MVS system, and it appears your LINE2 is set up for RMT02 not RMT1. Below is the TK4- configuration.
?
?RFEVIEW ?SYS1.JES2PARM(JES2PARM) - 1.02 ? ? ? ? ? ? ? ? ? ? ? ? ? Columns 1 72?
?Command ===> ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?Scroll ===> CS ?000149 *** ? ? ? * ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?000150 LINE2 ? ?UNIT=671,EBCDIC,TRANSP,HISPEED,FDUPLEX,NOADISC,CODEA,IFACEA ? ? ?000151 RMT2 ? ? 3780,NUMPR=1,LINE=2,NUMRD=1,NUMPU=1,TRANSP,NOTABS ? ? ? ? ? ? ? ?000152 R2.PR1 ? START,CLASS=XA,UCS=QN,FCB=6,SEP,AUTO,PRWIDTH=151 ? ? ? ? ? ? ?? ?000153 R2.PU1 ? CLASS=B ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?000154 R2.RD1 ? CLASS=A ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?000155 DESTID NAME=RMT02,DEST=R2 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? ?000156 *** ? ? ? * ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
I Hope This Helps
?
?... Mark S.
? |
The /*ROUTE RPINT RMT1 is intentional, to not crash the connection back
toggle quoted message
Show quoted text
to the VM/RSCS side. Otherwise, we see the IO errors (which is what I'm trying to resolve). On Thu, 2024-10-10 at 11:26 -0700, Mark A. Stevens via groups.io wrote:
On Mon, Oct 7, 2024 at 01:16 PM, Daniel L. Srebnick wrote:However, if I include a /*ROUTE PRINT RMT1 to send the output to myIIRC when you submit from a remote station (VM, or RJE) and you want |
On Thu, Oct 10, 2024 at 01:28 PM, Daniel L. Srebnick wrote:
Otherwise, we see the IO errors (which is what Look at the manual, and search for $HASP094, around page 98 of 416. There is an explanation of all the bits there.
?
http://www.bitsavers.org/pdf/ibm/370/MVS/GC28-1374-1_MVS_370_Message_Library_System_Messages_Volume_1_AHL-IEA_Jan85.pdf
?
$HASP-6 MVS/370 System Messages Volume 1
?
?... Mark S.
? |
Thanks for that pointer Mark...it is at least enlightening. ?Perhaps someone else can better interpret the message, but to me it looks like a read error on the JES2 side. On Thu, 2024-10-10 at 11:48 -0700, Mark A. Stevens via groups.io wrote:
|
Hi Daniel,
?
Try using the following for your RSCS link start command:
?
START NODEA PARM H02 B999
?
Page 35 in the manual "GC20-1816-3 VM/370 Remote Spooling Communications Subsystem Users Guide, Release 3" documents the parms for the start command and describes the H parm as the line number of the JES2 remote line VM is connecting to.? I'm suggesting B999 to bump the buffer size up to possibly counteract the I/O errors on data flowing from JES2 to VM.? I've used that on other links with success, so it may work here as well.
?
When I use this start command, I am able to start the RSCS link, followed by the JES2 line.? I get one I/O error when I start the JES2 line, but none after that.
?
I am able to use the RSCS CMD command to send commands to JES2 and get the response files back.? I am getting an "invalid destination" when I punch files to RSCS with NODEA in the tag, but I suspect that's because the AXSROUTE COPY file is empty.
?
Eventually, when I want to bring it down, I just $PLINE<n>.? When JES2 shuts down the line, RSCS detects that and deactivates the link.
?
Jeff
? |
Hi Jeff: I've gone both up and down in buffer size. I can send a short command or a short JCL from RSCS to JES. ?No problem there. ?But when I send a larger job, maybe 20 lines, the IO errors start. ?There seems to be something wrong in the way the BSC conversation over TCP is being handled. Now in the case of the short JCL (an IEFBR14 job, for example), it will run and output will come back to RSCS. ?Since the output is not tagged in any way to go back to my user, it gets printed on PRT1. ?Aside from the IO issues, the next thing I need to figure out is if there is any JCL that could tell RSCS to route the job back to PRT on my userid. On Sat, 2024-10-12 at 02:21 -0700, Jeff Snyder wrote:
|
I still can't punch jobs to MVS from VM for some reason but, in tests today, output routed from MVS to VM transfers with no I/O errors.? So far, I have routed jobs with various amounts of output, from 153 lines of MVS output (90 records on the VM SPOOL) to 11071 lines (11145 records on VM).
?
In case the issue is not with the RSCS start command, here are my Hercules config statements for the link.? They seem to mirror yours though, so I don't think there's an issue here.
?
Note: Pi4 is the hostname for my Pi host OS.
?
MVS - 670 is JES2 line 1, the one I used for connecting to VM
670 2703 dial=no lport=37908 rhost=Pi4 rport=37913 # RJE to VM/370 (JES line 1)
?
VM - 040 is the CUA defined as NODEA in RSCS
040 2703 dial=no lport=37913 rhost=Pi4 rport=37908 # RJE link to JS08 (NODEA - SML mode to MVS RJE line)
?
As for routing the output to a specific user, the best I've been able to come up with is draining the VM printer or changing the output class or whatever so the output doesn't print by default and using the TRANSFER command to move the output file to the requestor's queue.
?
JES2 thinks it's talking to a dumb RJE workstation, not a smart system with an OS and multiple users.? If you had a level of JES2 that supported NJE, you could use that and Peter Coghlan's updated RSCS to set up an NJE link and get closer to what it sounds like you want.? I've used his modified RSCS to support both NJE and RJE links with good success, though I don't remember specifically testing RJE jobs with massive output for I/O errors.? Typically just IEFBR14s to see the connection works.
? |
Here's an update.? After extensive testing and numerous suggestions from others in the community, I think I have found an answer.
?
On the VM/RSCS side, you need to start the line with a buffer size of 606.
?
START NODEA PARM H02 B606
?
Using this configuration I am able to cleanly punch jobs from a VM user, via RSCS, to JES2.? Unless I route the output otherwise, it comes back to the RSCS virtual machine, which prints the output on VM's PRT1.
?
I'm still testing, but am posting my results here in case anyone else wants to try. |
On Mon, Oct 21, 2024 at 11:11 AM, Daniel L. Srebnick wrote:
I'm still testing, but am posting my results here in case anyone else wants to try. Yes, I tried it and it works for me too. Thank you so much for finding this, I wanted that RJE connection to work for so long. :-)
?
Cheers,
?
Rene FERLAND, Montreal |
to navigate to use esc to dismiss