I have been working as a systems programmer on/with MVS and OS/390 for only 20+ years now, therefore, compared to quite a few people around, I am still a youngster (while not a rookie, though). It has always be my dream to have MVS running on my own computer, but obviously, my boss (aka as SO, aka as wife) would have seriously objected if I had tried to put some big blue (or red) hardware into our house. Imagine my surprise when I learned about Hercules some time ago... The chance to have a mainframe on your PC!
But... what is a mainframe without an Operating System. That is where Rick Fochtman came into the picture. He built a CD-Rom with OS/360 (MFT and MVT), and this CD I downloaded from the internet (well, the internet _is_ good for some things, once in a while....). I managed to get MFT and MVT operational on a spare computer running Linux, and I was hooked.
When finally Hercules was made available for Windows, and the MVS starter system became available, I seriously invested some effort into Hercules. This little document shows you how to create a working MVS 3.8 System on your PC, may it be Linux or Windows driven.
Please note that I have only Windows/NT, Windows 98, Suse Linux 7.0 and Red Hat Linux 7.0 available, and can only test my instructions on these test beds. If you are running something else, you are on your own... but these instructions should help you nevertheless.
You can save yourself a whole lot of work by using a pregenerated system. The easiest way is to install the Turnkey MVS system - You insert the CDRom, answer a few questions, and off you go into happy MVS land:
Insert the CDROM into a CDROM drive, change to the CDROM and enter ./setup. A
simple shell script will start and display
+=================================================== ! ! MVS 3.8J System Installation/Generation ! ! 1. Install MVS Turnkey system ! 2. Install MVS Starter system and SYSGEN files ! x. Exit ! +=================================================== Your choice ===>You now select either
1 if you want to install a ready to run MVS system, or
you select 2 if you want to generate your own system.
If you are running this script under Windows, you will see next:
This is CYGWIN running under some flavo(u)r of Windows Please enter the drive letter of the CDROM drive or press enter for default (D)You must enter the drive letter of the CDRom drive. Note that this script does not try to autodetect the CDROM drive, it just assumes D as the default drive letter if you don't provide it. This prompt will not appear if running under Linux, as Linux doesn't have the concept of drive letters.
The following prompt will also not appear under Linux , but under Windows:
Please enter the target drive letter, or press enter for default (C)Here you determine onto which drive the mvs turnkey system should be installed
The next prompt will be similar for Linux and Windows systems:
Please enter the target directory, or press enter for default (mvs)
Note: The directory will be used relative to the root directory
Use a forward slash (/) as a directory indicator
Do _NOT_ use a backslash (\)
do _NOT_ use a / at the beginning of the path
With other words, if you just press the enter key, the turnkey MVS system will be installed
in /mvs on the root directory (linux) or the target drive (Windows)
Remember to use a forward, not a backward slash, even in Windows. With other word, if you
want to install into directory mvs under the directory bigfun
you would now enter
bigfun/mvs
Another prompt appears.
Default installation directory will be used: mvs
Turnkey Installation Option
1. Use DASD images from CD-ROM
(Somewhat slower, requires about 30 MB disk space)
2. Copy DASD images to local hard disk
(faster, but requires about 260 MB disk space)
Enter your choice (1/2) ===> 1
Here you have the coice between space and speed. If you choose option 2, the DASD
image files will be copied to your hard disk, which will take a few minutes
When copying is complete (or has been skipped), you will now see
Creating directory d:\mvs Do you want to copy the pre-installed version of Hercules 2.12 for Windows that is on this CD-ROM to your harddisk? (Y/N): [N]There is a Hercules version provided on the CDROM. You might have hercules installed already, then you wouldn't want to use the supplied version. If, on the other hand, you don't have hercules installed, or have an older version, then you might want to reply
Yto the prompt. On a Windows system, the precompiled version will just be copied to a directory onto your hard disk. On a Linux system, the Hercules Version supplied will be compiled, linked, and copied into your /usr/bin directory. If, on linux, you do not want to have the hercules binaries placed into /usr/bin, you answer "N" to the prompt, and then manually install hercules via
cp [cdrom]/hercules-2.12 [hercules-dir]
cd [hercules-dir]
{change makefile to point to your target directory}
make HOST_ARCH=`uname -m` install
When hercules is installed (or the installation is skipped) a few more messages are displayed, and finally you see
Setup is now complete
You can find the mvs installation in directory d:\mvs
To start MVS, change to that directory and invoke
startmvs
To start ZZSA, the Stand Alone editor, invoke
startzzsa
Press the ENTER key to leave the setup program now
Change to the directory where you installed mvs, run the startmvs
script.
In order to really USE your system, you will need a tn3270 client. I did include a windows client, the QWS3270 free client. This will work as a master console, but has serious drawbacks when used in a TSO session.
You can download another tn3270 client from the x3270 homepage. It comes in two flavours, a graphical version and a text version. I have included the a text-mode tn3270, called c3270. It is contained in the directory c3270-3.2. In order to build it, change to that directory, and then enter
./configureand then
makeThe CD-Rom also contains the documentation in the c3270-3.2/html subdirectory. Note, that you need two tn3270 sessions for OS consoles, and one to n client sessions for TSO sessions. These must be started before you IPL the system.
A standard telnet client is also needed as a safety measure. In a Windows/NT environment the client will be automatically started from the hercules.rc file in the conf directory. The telnet client needs to be manually started in other environments.
There seems to be a serious problem with Telnet under Windows 2000. As an alternative you may wish to use Hyperterm, which is a communication program that can use TCP/IP to connect to our system as well.
You can now ipl the system off devive 148 (volume MVSRES)
If you are not quite sure on how to do this IPL thingie, take a look at How to start MVS
If you need some brush-up on MVS and JES2 commands, take a look at the cheat sheet
You should read the instructions for building the whole system in order to find out which features are available.
A Few Highlights:
HEL ZAP to see
the help text for the ZAP commands. You can use HEL UTILS to see which utilities
have been installed into your version of the MVS turnkey system
Currently, this distribution is only available on CDROM
+ installation directory ! +------ conf configuration and control files ! +------ dasd pregenerated dasd images ! +------ herclin directory for Hercules binaries for linux ! +------ hercwin directory for Hercules binaries for windows ! +------ jcl contains the SYSGEN Job Control Files ! +------ scripts command scripts used during installation ! +------ snapshot Snapshots made during sysgen ! +------ standalone MVS standalone programs ! +------ tapes Directory for tape files ! +------ xmi MVS PDS files in xmi format
When you need to run command files, you will find them in the "scripts" directory. The command files for windows have the extension ".CMD", while the Linux scripts have now extension at all.
If you have a newer version of Hercules (Linux or Windows) feel free to place it into the correct subdirectory (hercwin or herclin)
Note that the Hercules version must have support for compressed disks and tapes using the bzip compression algorithm included, as all files used by this generation are compressed that way.
If in doubt, use the Hercules version provided in this package, it has the necessary compression enabled
Now the real fun starts. You first have to build a starter system. Don't worry, this is really not difficult. Just run the Installation script
This scripts unpacks the starter system from an archive provided in the "snapshot"
subdirectory. No error messages are expected from this script.
Table of Contents
The ins0020 builds two 3350 disk image files, one for the system residence, and one for the DLIB volume. Be patient, this will take a few moments
After the command script has finished, you should find two new files in the dasd directory
Note that the mvsres.148_n is not compressed, as opposed to mvsdlb.248. The compression utility only works on devices that do not have alternate tracks specified, while the utility that write IPL text requires alternate tracks. We therefore compress the system residence volume AFTER we have placed the IPL text on it.
Invoking the command script ins0030 under Windows will start your telnet client, and then the Hercules engine. Under Linux, only the Hercules engine will be started. In order to fire up the telnet client, you need to switch to another linux session and invoke the starttelnet script (which starts the telnet client).
Make sure that your telnet client connects to port 3270 on the machine where Hercules is running. If this the same machine as where the telnet client will be acivated, you can use the following command line:
telnet localhost 3270
Once the Hercules engine is started, and the telnet client has connected to it, you can now IPL the stand alone init utility. On the Hercules console you enter
ipl 00aor, on the semi-graphical console that you reach by pressing the ESC key you enter
l bAfter a short while the system should enter an enabled wait. Switch over to the telnet console and press the >enter< key once You will see something like:
IBC105A DEFINE INPUT DEVICE. DASDI 7.92 HHC901I Enter input for console device 0009to which you answer (by typing at the telnet console):
input=1442,00cand again pressing the >enter< key once. The telnet console should now show:
MVSRES JOB 'INITIALIZE 3350 MVSRES VOLUME'
MSG TODEV=1052,TOADDR=009
DADEF TODEV=3350,TOADDR=148,IPL=NO,VOLID=MVSRES,BYPASS=YES
VLD NEWVOLID=MVSRES,OWNERID=HERCULES
VTOCD STRTADR=1,EXTENT=29
END
Be patient, the system is now initializing the disk pack, and that will take
a few minutes. When you see the message
IBC163A END OF JOB. IBC202A INTERV. REQUIRED 00C 02 0E00 4000 0000 0000 IBC154A READY READER 00C. DEPRESS INTERRUPT KEY.then the disk is initialized, and you can now terminate the Hercules engine and the telnet client
Invoking the command script ins0040 will now (again) start the Hercules emulator, but with a different configuration file. This time, after the telnet client has connected, you will want to start the MVS starter system by entering
ipl 150on the Hercules console!!. After a few seconds, the telnet session should show
IEA101A SPECIFY SYSTEM PARAMETERS FOR RELEASE 03.70.VS2 HHC901I Enter input for console device 001FNow you enter
r 00,clpabecause this is the first IPL of the starter system. If you start the system at some later time, you can just press the >enter< key as a CLPA is not generally required then
After about a minute or two you should see
IEF165I // START JES2 IEE351I SMF SYS1.MAN RECORDING NOT BEING USED *00 $HASP426 SPECIFY OPTIONS - HASP-II, VERSION JES2 4.0To which you reply
r 00,format,noreqAgain, this is only required because this is the first time that JES2 gets started, and the spool dataset SYS1.HASPACE needs to be formatted. If you start the system later, you would enter
r 00,,noreqBack to our IPL, though. A message will now be displayed
*01 $HASP436 REPLY Y OR N TO CONFIRM CHECKPOINT RECORD CHANGEto which your reply is
r 01,yAgain, you need to be patient, JES2 is now formatting the SPOOL0 volume. When everything has been formatted, you will see
IEE041I THE SYSTEM LOG IS NOW ACTIVE $HASP100 INIT ON STCINRDR $HASP100 INIT ON STCINRDR $HASP100 INIT ON STCINRDR $HASP373 INIT STARTED $HASP309 INIT 1 INACTIVE ******** C=A $HASP309 INIT 2 INACTIVE ******** C=BA $HASP309 INIT 3 INACTIVE ******** C=CBA $HASP373 INIT STARTED $HASP373 INIT STARTED $HASP099 ALL AVAILABLE FUNCTIONS COMPLETEAlso, Hercules will now feed a job with jobname sg0000 to the started JES.
You should now enter the following JES commands
$SPRT1 $SPUNCH1to enable printing/punching of job output.
After the job sg0000 has completed, you will find the output in the prt subdirectory with the name sg0000.lst
| Expected RC |
|---|
| 0 |
The JCL to be run for the system generation are in the jcl subdirectory. To submit a in file sg1234 you have two possibilities:
Sometimes it is nice to see the JES2 assigned job numbers when diplaying MVS information via operator commands. You can achieve this by entering
$T OSC1,D=JIf you would also like to see a timestamp on every message, then enter
$T OSC1,D=Tinstead
You might also consider to issue the MVS command
mn jobnames,twhich tells MVS to monitor and report on start and end of jobs.
All jobs that need to be run on the starter system have names beginning with sg followed by a number. Other jobs in the jcl directory will be run from the generated MVS system later
The SMP4 that is distributed with MVS is unfortunately unfit for system generation!! Fortunately, a replacement is available. This job will copy it from a work file into the starter system's SYS1.LINKLIB.
While we are updating some system files, we also add a few procedures to the system procedure library
| Expected RC |
|---|
| 0 |
As already mentioned an IPL is now required. No, don't press ctrl-alt-del yet! At the telnet screen enter the commands
$p $pjes2 z eodThen, at the Hercules console, enter
stop
We could now do an IPL. But, this time, we actually want to leave Hercules altogether, as we can now take the opportunity to compress the system residence. Leave Hercules by entering the command >exit< on the Hercules console, or by using the >W< command on the graphical Hercules console
Once Hercules is terminated, you will now have to run
The script ins0050 will compress the mvsres system residence volume, but, just
in case, will also place a backup copy of the uncompressed file into the snapshot subdirectory.
The compressed file will have the name mvsres.148, and the archive with the uncompressed file
will have the name mvsres_148_not_compressed.tgz.
Table of Contents
The script file ins0060 has a twin, called startstarter (easier to remember). It can be used to start the MVS starter system. Remember that you will have to start a telnet client on port 3270 (using, on Linux, the tn script or the starttelnet script)
The MVS system to be generated needs a mastercatalog. We will place it on the system residence, and connect it to the starter system as a user catalog. Just run job sg0020 and see what happens...
| Expected RC |
|---|
| 0 |
The DLIB (=Distribution LIBrary files) contain the building parts out of which the target system (the MVS to be generated) will be built. We allocate empty DLIB files using job sg0030. Later, these files will be filled via SMP4. Therefore we need to initialize the SMP environment as well.
| Expected RC |
|---|
| 0 for all but last step |
| 8 for last step on first run |
| 0 for last step repeated |
Now we have to feed the product tape to SMP, which will unload the parts from the distribution tape and place it into temporary libraries on the MVSRES volume. (That is the volume that we specified on the TLIB parameter of the SMP4 procedure in job SG0010).
The distribution tape is in the file zdlib1.het, and you load the tape drive by entering
devinit 170 pathname/zdlib1.hetwhere pathname is the path to the directory containing the tape file. If you kept the the file in the tapes subdirectory the command would be
devinit 170 ../tapes/zdlib1.hetNow submit job sg0040. Be patient, this job has a lot to do, and it takes quite some time to complete!
| Expected RC |
|---|
| 0 |
Okay, so the long wait is over. What are we going to do now? Well, it is
time for a snapshot.
Table of Contents
You have invested quite a big amount of time and effort so far. Wouldn't it be a shame to loose it all, just because, like, by accident you reran command file ins0010? Better safe than sorry, thus let us create a backup.
First shut down MVS and end the Hercules engine. Then invoke command file ins0070 which will use TAR and gzip to create several files in the directory snapshot.
A volume can be recovered from a snapshot using command script restsnapvol. The syntax is
restsnapvol {snapshot} {volume}
! !
! +--- volume of disk to restore
+-------------- snapshot number (01, 02, etc)
Example: restsnapvol 01 mvsres
to restore volume MVSRES from snapshot # 1
A complete snapshot can be restored command script restsnap. The syntax is
restsnap {snapshot}
!
+-------------- snapshot number (01, 02, etc)
Example: restsnap 01
to restore the complete snapshot # 1
After the snapshot is complete, you can now restart Hercules, the telnet client, and MVS
No software product is without bugs. This is (especially) true for MVS. In the course
of time fixpacks etc have been developed. IBM call the fixes PTFs. Ever wondered,
what PTF stands for? Well, PTF = Program Temporary Fix
Okay, some of the fixes are available to us. A Tape with accumulated PTFS is available, with the name of CUM003. Submit the jog sg0060, and a mount message will promptly appear on the MVS console. If you placed the AWS file into the tapes subdirectory (as recommended), you can now load the tape drive from the Hercules console by entering
devinit 170 ../tapes/cum003.het
The maintence on the tape will now be received. But wait - the job doesn't end, it asks for another tape via a mount message. What do you do now? Very simple: On the Hercules (Hercules, not MVS) console enter
devinit 170 ../tapes/uz61013.hetSoon, there will be a new mount message, to which you reply
devinit 170 ../tapes/uz71613.hetThere will be a last mount message, and you supply
devinit 170 ../tapes/mvsptf1.hetAll the maintenance currently available is now placed into the SMP environment.
| Expected RC |
|---|
| 8 |
Note that this job will have a return code of 8, as some PTFs appear twice, and SMP will record this as an error.
On the MVSDLB volume there is a file called SYS2.SMPPTS which contains maintenance form a similar system, but from some time later. We will RECEIVE this maintenance in job sg0060 to have those PTFs available should we ever need them. A return code of 0 is expected for this job.
| Expected RC |
|---|
| 0 |
For some compelling, SMP related reason it is necessary to install one specific product before all the others. Install in this moment means: Load the information from the SMPPTS and the SMPTLIBS (where they have been placed by the RECEIVE commands in SG0050-SG0080) onto the DLIB files on MVSRES.
This will be done using the ACCEPT command of SMP. Usually, though, a SYSMOD will not be ACCEPed right away, but needs to be put into the Target Sytem for testing first, which would be done via an APPLY command. But we do not have a target system in the moment. Therefore the parameter NOAPPLY on the ACCEPT command is required to tell SMP that the FMID should be ACCEPTed even though it has not yet been APPLYed.
If you don't want your SMP jobs to run forever, use DIS(WRITE) as an additional option on the ACCEPT command. This keeps the directory of the highly used PDS SYS1.SMPACDS in memory, instead of incurring gazillions of disk I/O operations. If things go wrong, though, your SMPACDS might be seriously broken... Guess why we did a Snapshot in the previous step!
| Expected RC |
|---|
| 0 |
The next job will install the BCP (=Base Control Program) plus a lot of related FMIDs (Functions) needed by MVS. This is a rather big job, and on my system (500 MHz, 256MB) it took about 3hours 20 minutes. Thus, have a break, watch the blinkenlights, and be patient
I also ran the same job under Linux on a 150MHz, 80MB Thinkpad. It was not exactly fast.....
| Expected RC |
|---|
| 0 |
After such a massive change to our system it might be a good idea to take another Snapshot. Shutdown MVS and the hercules engine, then run command script ins0080, which puts new files into the snapshot directory. Don't throw away the previous snapshot files - you never know....
So far, everything has been rather straight forward. Now, we come to a difficult spot. Some of the FMIDs to be installed now may refuse to be installed, because a few IFREQ PTFS may be missing. This can be healed by also ACCEPTing UY49086, but this (at least on my system) breaks SVC94 (which is used for GTSIZE et al) and makes part of TSO (like TEST) unuseable. I do not know which PTF fixes UY49086, don't have the means to find out, and (probably) don't have the fixing PTF either. UY49086 also modifies SYSGEN for TCAM incorrectly (it will introduce a global variable &SGDTCAM that is referenced nowhere else). Instead of supplying the faulty PTF I leave it off altogether by specifiend BYPASS(IFREQ). If you prefer you can add UY49086 to the list of SYSMODS instead (and live with the consequences)
| Expected RC |
|---|
| 4 |
This one is really hairy. There are about 29 missing PRE or IFREQ SYSMODS, and there is nothing we can do about it right now. We will install the processor support group with BYPASS(IFREQ,ID) option to get a somewhat clean install. Expect a return code of 4 because of the BYPASS option.
| Expected RC |
|---|
| 4 |
The Job entry subsystem is straight forward in terms of installation. Just run job sg0110 and expect a return code of 0.
| Expected RC |
|---|
| 0 |
We have completed the first phase of the sysgen process, the DLIB disk is now loaded. This is a perfect time for another snapshot. With other words, shut down MVS and the Hercules engine, and then run command script ins0090
The DLIBs are loaded, and we need room on the disk packs for the target libraries. We use SMP/4 to do some of the cleanup for us. If you run job sg0120 this will REJECT all the function SYSMODs. For ACCEPTed function SYSMODS this means that the SMPTLIB files (which were allocated on MVSRES) will be removed. Also, this job will run IEHPROGM with the SCRATCH VTOC function to remove all non-VSAM files from the MVSRES volume. You shold expect a return code of 4 for the first step, as the rejected function SYSMOD have not been APPLYed. There is a second step which cleans the MVSRES pack via a SCRATCH VTOC operation. You will get a RC=8 for this step, which is okay.
| Expected RC |
|---|
| Step1 : 4 |
| Step2 : 8 |
The SYSGEN process needs some datasets, which will be allocated in the job sg0130. Also, a new jobcard macro will be placed in SYS1.AGENLIB which includes a JOBCAT DD statement.
| Expected RC |
|---|
| 0 |
Sometimes the assembler provided by the starter system has the hiccups and randomly abends with S0C1 or S0C4. We have all the parts to build a new assembler, and that is exactly what SG0140 is doing. After this job has completed successfully (we expect RC=0) then you must shutdown MVS and Hercules! If you forget this, various abends (like S106-0E) may occurr.
| Expected RC |
|---|
| 0 |
We will use two dedicated page volumes (3340) for our target system. Shutdown MVS and the Hercules engine if you have not done so already. When Hercules has terminated, run command script ins0100, and two disk files (page00.160 and page01.161) will be added to your dasd subdirectory. Also two lines will be added to the end of the starter.conf configuration file
After this command script has completed, you should start Hercules and then IPL MVS again.
| Expected RC |
|---|
| 0 |
Finally! There are quite a few comments in the stage1 assembly deck, and I would recommend that you take a look at the deck before running job sg0150.
The IO Configuration can be viewed by clicking on one of the following links:
| Expected RC |
|---|
| 0 |
The stage 1 of the previous job step has placed a large JCL stream into file SYS1.STAGE1.OUTPUT. Now go to the Hercules console and enter
devinit 13 ../jcl/stage2.pch asciibefore you run the job sg0160. This will place the JCL stream into file stage2.pch in the JCL subdirectory. In order to make this file available for editing, enter
devinit 13 ../pch/pch00d.txt
| Expected RC |
|---|
| 0 |
Some modifications need to be done before the stage2 jobs can run. The easiest way to do this is running the script ins0110 which does the changes for you. If you want to see which changes are being done, take a look at conf/stage1.sed
Do not run this/these jobs unless you really have to!!!
Sometimes, you need to start a sysgen stage2 process all over again. Then you might want to delete the target libraries, first. This is accomplished by jobs sg0170 for the full sysgen, and jobs sg0170_x (1 <= x <= 5) for sysgen jobs SYSGENx. Remember, you only need these jobs when restarting (a part of) sysgen process!!
The script ins0110has created the jobstream
The real Stage 2. It is a sequence of 6 jobs to build the various MVS target libraries.
Note: The stage1 macros specify class C as the jobclass where the sysgen jobs are to run. The starter system has 3 started initiators, where only one has class C. This assures that the jobs run consecutively, not parallel
at the Hercules console enter the following command:
devinit 012 ../jcl/sg0180 trunc eof(don't forget the trunc). Then take a break, because it will take some time to run the sysgen jobs.
Now everything is in place, nearly. We still have to build a JES system, but that is quite simple. Run the job sg0190, which Links JES2, sets up a JES2 procedure, and loads some simple JES2PARMS.
The job will also allocate a (rather small) spool dataset on MVSDLB. We will later expand the system by dedicating the better part of a full volume to the spool, but we will wait until we are certain that our generated system will start successfully.
| Expected RC |
|---|
| 0 |
A few changes are needed to SYS1.PARMLIB members. These changes will add SYS1.CMDLIB to LNKLST00, select a different IEAPAK00, etc. Run job sg0200 to implement the changes. After this job has completed, shutdown MVS and the Hercules engine.
| Expected RC |
|---|
| 0 |
The actual System Generation is done (even though we need to do some customizing, still),
and we should now do another snapshot (by running script ins0120.
Table of Contents
This is the first time MVS 3.8 is ipled, therefore an MVS CLPA is required, and later a spool formatting JES2 start.
During IPL with CLPA you will see the following messages:
| IEA101A SPECIFY SYSTEM PARAMETERS FOR RELEASE 03.8 .VS2 | r 00,clpa | IEA357I ERRORS IN IEAPAK00 LIST | IEA357I IGC0J05B INVALID - MODULE IGNORED | IEA357I IGG08101 INVALID - MODULE IGNORED | IEA357I IGG08102 INVALID - MODULE IGNORED | IEA357I IGG08103 INVALID - MODULE IGNORED | IEA357I IGG08104 INVALID - MODULE IGNORED | IEA357I IGC11110 INVALID - MODULE IGNORED | IEA357I IGC12110 INVALID - MODULE IGNOREDThese are no errors, just information, and you can safely ignore them. After MVS is started there are three printers in the the system, at addresses 00E, 00F, and 30E. 30E is generarted as a hardcopy device while 00E and 00F are standard printers.
Issue the JES2 command
$T prt2,Q=Zto make the second printer serve the output queue Z. This might be handy sometimes later.
At some time during startup you will see a message similar to
*$HASP190 DUMMY SETUP -- PRINTER1 -- F = STD. -- C = 6 -- T = 0When this setup message appears, you must start the corresponding printer, in this case by entering
$S PRINTER1
Having fun with your MVS? Well, we have some more work to do!
The basic system generation is complete, but we are no ways done yet. There are a lot of things left to do until the system is useable, like adding disk space, making TSO available, installing an editor and other utilities, etc.
The current system is a two volume system, where there is not much free space left for expansion. One of the consequences is that our spool dataset will fill up fairly quickly. Also, we will want to have work files, permanent storage for TSO users, Sort work areas, etc. With other words, we will need more disk space. Fortunately, this is not very complicated.
VAT stands for Volume Attribute List and allows automatic mounting of new disk volumes. We will create new disk volumes soon, therefore you should already run job mvs0010 to place a new VATLST00 into your SYS1.PARMLIB library.
| Expected RC |
|---|
| 0 |
Shut down MVS and the Hercules engine now. Then run
Table of Contents
This script will build quite a few DASD image files in your ../dasd subdirectory. When it is complete, you will find the following
The script will also update the MVS.CONF file used for Hercules startup, such that the newly created disk files can be accessed by MVS.
Now you should start Hercules again, and then MVS. After MVS has completely initialzed, and JES2 was started, enter the MVS command
D U,DASD,ONLINEon the MVS master console. You should get a result similar to this:
d u,dasd,online IEE450I 16.17.17 UNIT STATUS 086 C UNIT TYPE STATUS VOLSER VOLSTATE UNIT TYPE STATUS VOLSER VOLSTATE 130 2314 O SORT00 PUB/RSDNT 131 2314 O SORT01 PUB/RSDNT 132 2314 O SORT02 PUB/RSDNT 133 2314 O SORT03 PUB/RSDNT 134 2314 O SORT04 PUB/RSDNT 135 2314 O SORT05 PUB/RSDNT 140 3350 O WORK00 PUB/RSDNT 141 3350 O WORK01 PUB/RSDNT 148 3350 S MVSRES PRIV/RSDNT 150 3330 O SPOOL0 PUB/RSERV 160 3340 A PAGE00 PRIV/RSDNT 161 3340 A PAGE01 PRIV/RSDNT 240 3350 O PUB000 STRG/RSDNT 241 3350 O PUB001 STRG/RSDNT 248 3350 A MVSDLB PRIV/RSDNTwhich shows you the accessible volumes and their mount attributes.
We will make immediate use of the new volumes. Let us first make a dedicated Spool pack:
The spool space on MVSDLB is very small, and we would run into problems if we wanted to use it in "real" work. This is where mvs0020 will help you - When this job is run, it will
| Expected RC |
|---|
| 0 |
STC 7 $HASP100 NET ON STCINRDR
STC 7 IEF452I NET JOB NOT RUN - JCL ERROR
- STC 7 $HASP396 NET TERMINATED
STC 7 $HASP150 NET ON PRINTER2 7 LINES
*STC 7 *$HASP190 NET SETUP -- PRINTER2 -- F = 0001 -- C = 6 --
* T = QN
STC 8 $HASP100 TSO ON STCINRDR
STC 8 IEF452I TSO JOB NOT RUN - JCL ERROR
- STC 8 $HASP396 TSO TERMINATED
Don't worry, there is nothing wrong with your system. There are commands to start
VTAM and TSO, but we haven't actually setup those two components yet (but will do so shortly)
Which changes were implemented to JES2PARMS? You might want to look at the member in SYS1.PARMLIB. The main changes are:
The new spool is working, we now can remove the spool space from MSDLB. This is done by running job mvs0030.
| Expected RC |
|---|
| 0 |
We are now ready to build VTAM and TSO:
Just run this job. When it is completed, some new files have been created, and some members in SYS1.PARMLIB have been updated. The following files were created:
Following procedures have been added to SYS2.PROCLIB:
| Expected RC |
|---|
| 0 |
A straight forward affair. Just run this job, and it will create Libraries and LOGON procedures for you:
logon someoneto logon to TSO using the userid "someone"
| Expected RC |
|---|
| 0 |
which will assemble Logon Interpret Table and logmoe table.
| Expected RC |
|---|
| 0 |
Go to the MVS console, and enter
s net
IF TCAM was not included in the Stage1, you should see the messages
- STC 19 $HASP373 NET STARTED
- STC 19 IEF403I NET - STARTED - TIME=17.44.23
STC 19 IST025I BLDL FAILED FOR ISTYTCM1 IN VTAMLIB
STC 19 IST126I ISTYTCM1 MODE NOT SUPPORTED DUE TO LOADING FAILURE
STC 19 IST025I BLDL FAILED FOR ISTYTCM2 IN VTAMLIB
STC 19 IST126I ISTYTCM2 MODE NOT SUPPORTED DUE TO LOADING FAILURE
STC 19 IST025I BLDL FAILED FOR ISTYTCM3 IN VTAMLIB
STC 19 IST126I ISTYTCM3 MODE NOT SUPPORTED DUE TO LOADING FAILURE
STC 19 IST110I NETWORK SOLICITOR STARTED
STC 19 IST093I APPLTSO ACTIVE
STC 19 IST093I APPLCICS ACTIVE
00 STC 19 IST093I LCLMAJ00 ACTIVE
STC 19 IST020I VTAM INITIALIZATION COMPLETE
IF TCAM has been included, you will see:
- STC 19 $HASP373 NET STARTED
- STC 19 IEF403I NET - STARTED - TIME=17.44.23
STC 19 IST025I BLDL FAILED FOR IEDIAE IN VTAMLIB
STC 19 IST025I BLDL FAILED FOR IEDIAE IN VTAMLIB
STC 19 IST025I BLDL FAILED FOR IEDIAE IN VTAMLIB
STC 19 IST025I BLDL FAILED FOR IEDIAE IN VTAMLIB
STC 19 IST025I BLDL FAILED FOR IEDIAK IN VTAMLIB
STC 19 IST025I BLDL FAILED FOR IEDIAK IN VTAMLIB
STC 19 IST110I NETWORK SOLICITOR STARTED
STC 19 IST093I APPLTSO ACTIVE
STC 19 IST093I APPLCICS ACTIVE
00 STC 19 IST093I LCLMAJ00 ACTIVE
STC 19 IST020I VTAM INITIALIZATION COMPLETE
These messages relate to TCAM, which wasn't installed or generated (yet), and can be ignored
When VTAM is initialized, enter
s tsoto which the system should reply
- s tso
STC 20 $HASP100 TSO ON STCINRDR
- STC 20 $HASP373 TSO STARTED
- STC 20 IEF403I TSO - STARTED - TIME=17.46.08
00- STC 20 IKT007I TCAS ACCEPTING LOGONS
- STC 20 IKT005I TCAS IS INITIALIZED
After the message TCAS IS ACCEPTING LOGONS appears, you can now
logon using the IBM supplied userid IBMUSER. How? Go to a 3270 session that displays
THIS TERMINAL IS LOGGED ON TO THE NETWORK SOLICITORand in the top row enter
ibmuser size(4096)Note: The default size for IBMUSER is 44K, which is enough for running ACCOUNT, but not for running anything else. If logon worked, just enter LOGOFF again. We are going to create new TSO users in a minute
Note: When you are not sure about the Syntax of VTAM commands, take a look at the the VTAM 2 Commands Cheat Sheet
We do not want all the files for specific TSO users to be cataloged in the MVS master catalog. Therefore we build a usercat on volume PUB000 for all the TSO related files. Run job mvs0070
| Expected RC |
|---|
| 0 |
We need a file containing skeleton for commands to make certain tasks (like creating users) easier. Job mvs0080 builds two files for use in a TSO environment:
| Expected RC |
|---|
| 0 |
You might want to modify mvs0090 before running it to suit your needs. AS written it will add four users to the system
| Expected RC |
|---|
| 0 |
It is time for another snapshot. We need to shutdown MVS again, but this time we first have to stop tso and net.
At the MVS console enter the command
p tsoif there are still users active, you will see a message like
00- p tso
*STC 22 *03 IKT010D 00001 USERS ACTIVE, REPLY 'SIC' OR 'FSTOP'
to which you reply
r xx,FSTOPif there are no (more) active users, you will see the message
*STC 22 *04 IKT012D TCAS TERMINATION IN PROGRESS - SPECIFY 'U' OR 'DUMP'
to which you need to reply
r xx,uTSO will now terminate. You can now terminate VTAM with one out the the following three commands
z net z net,quick z net,cancelEventually, you will see the message
STC 21 IST102I VTAM IS NOW INACTIVE
- STC 21 IEC223I IFG0200V,NET,NET,UNKNOWN
00- STC 21 $HASP395 NET ENDED
and now you can terminate JES in the usual way
$P $PJES2 Z EOD QUIESCE
This snapshot will NOT save all the dasd image files you have. It will only save
When you now start Hercules, and MVS, you will possibly see a message
*STC 31 *01 IKT003D TCAS UNABLE TO ACCEPT LOGONS, REASON CODE=0092 REPLY * 'RETRY' OR 'TERM'This happens because TSO starts up faster than VTAM. When VTAM has initialized, enter
r 01,retryand then TSO will connect to VTAM.
This jobs builds a few procedures in SYS1.PROCLIB fro SMP4 processing:
Just run this job, no modifications are necessary, and no problems are expected.
| Expected RC |
|---|
| 0 |
In order to use SMP/4 on the target system we need the SMP/4 target libraries. The job mvs0100 will allocate the files, and copy the contents of the DLIB SMP libraries into the newly created files.
Note: This job takes a long time, and uses quite a bit of CPU cycles. In order to avoid S322 abends on somewhat slower machines, I have added a TIME=1440 parameter to JOB and EXEC JCL statements
| Expected RC |
|---|
| 0 |
The stage2 job contains the building instructions for making our MVS system. Performing a JCLIN process as in mvs0120 will feed these instructions into SMP/4. SMP will then be able to rebuild elements of the MVS system
| Expected RC |
|---|
| 0 |
To test the SMP/4 environment we will install a simple usermod which will authorize some external commands to TSO. We will install these commands later, for now we will be happy when the usermod gets correctly installed. Just run the job mvs0130 and check that the module IKJEFTE2 gets linked correctly.
| Expected RC |
|---|
| First time: 12 |
| If repeated: 4 |
Why don't we get RC=0? Very simple, in order to make it possible to rerun this job, I put in a command to REJECT the usermod first. Obviously, there is nothing to reject, that gives the RC = 12. If the job is later rerun, then the REJECT will work, but gives a wrning (RC=4) because the usermod being rejected has not been ACCEPTed.
Note: Don't ACCEPT this Usermod!!
For the next activity we need some private macros. Just run this job to load the macros into SYS2.MACLIB
| Expected RC |
|---|
| 0 |
Well, this was simple. Now to a more complex adventure: A usermod that relinks the master scheduler. If this works, we can be rather sure that our sysgen process was successful. The job mvs0150 will assemble the element IEFACTRT and will ask SMP/4 to put it into our MVS. After the job has finished, SMP/4 should have relinked module IEFW21SD.
| Expected RC |
|---|
| First time: 12 |
| If repeated: 4 |
Why don't we get RC=0? Same reason as above. In order to make it possible to rerun this job, I put in a command to REJECT the usermod first. Obviously, there is nothing to reject, that gives the RC = 12. If the job is later rerun, then the REJECT will work, but gives a wrning (RC=4) because the usermod being rejected has not been ACCEPTed.
Note: Don't ACCEPT this Usermod!!
What does IEFACTRT do?
It will put a job step completion summary in the top of
the job log. No need for hunting for the many COND CODE lines any more....
Also, IEFACTRT does some step accounting. I used some numbers from 20 years ago to give you an
idea how much you would have to pay for running MVS jobs if this wouldn't be your
private PC.....
You will see its action after the next IPL with the
CLPA option. Therefore - shutdown TSO, NET, MVS and re-IPL with CLPA
Table of Contents
When user log off from TSO, you will see an error message IEC233I for IGV0200V. Peter van der Woude has located an error resolution in the IBM service databases. The relevant PTF is unfortunately no longer available, but he copied the information from the APAR AZ58122 and made it availble. The job mvs0160 will use SMP/4 to install this fix. After installing it, an IPL with CLPA is required.
| Expected RC |
|---|
| 0 |
MVS is up and running, the network is up, TSO is working - what more do we want? Well, there are dozens of little helper programs that can make your life easier when using TSO. Many of these can be found on the so called CBT-Tape and/or CD-ROM. They are available in source code, and you might need to customize them before being able to use them.
The tso0010 job creates a file SYS1.TSOCP.ASM to contain source for TSO command processors, and two procedures in SYS2.PROCLIB:
| Expected RC |
|---|
| 0 |
The job tso0010 assembles and links a tiny TSO command, called CLRSCRN. It can be used to clear the screen of a 327x display terminal (Don't ask me why IBM hasn't provided such a command as part of the TSO package - they just didn't!). Run this job to see if the command gets compiled correctly, and then log on to TSO to test it. Oh, CLRSCRN is too much of typing for you? The command has more than one name, pick the one you like most:
| Expected RC |
|---|
| 0 |
The job tso0030 loads a few TSO utilities from tape into SYS2.CMDLIB. When you run it, you must provide the tape file BSP001.HET (which is located in the tapes subdirectory). The following commands will be installed (together with help members in SYS2.HELP:
These programs will be loaded into SYS2.CMDLIB, and the corresponding help files will be in SYS2.HELP. You can use the TSO HELP command. or the fullscreen HEL command, to read the help info. Try
HEL PDSAAto see the help text for the extremely useful PDS command processor
| Expected RC |
|---|
| 0 |
The job tso0040 installs the VSAMWAIT utility into SYS2.LINKLIB and alters the installed TSO procedure. When you run it, you must provide the tape file BSP003.HET (which is located in the tapes subdirectory). On the next IPL you will no longer have to reply to the TSO prompt when TSO starts faster than VTAM.
In order to use some of the TSO utilities we just loaded we need an SVC set makes the environment authorized. I have provided a tiny SVC 244 for this purpose. Run job tso0050 to implement this user SVC. Note that an IPL with the CLPA option is required before the SVC becomes available
One thing is dearly missing in our MVS system: SPF. Unfortunately, SPF is a licenced IBM software product, and there is little chance of legally getting a copy of it to run on MVS 3.8 and Hercules.
But there is a silver lining on the horizon: Rob Prins has written a full screen editor that is somewhat SPF alike.
There is a
RPF User's Guide in your doc directory.
RPF is
rather easy to implement:
Table of Contents
This job will build some files (including the RPF source) on the PUB000 volume. Read the JCL and the comments and make all the changes needed for your system, then run the job. It will need the RPF product tape, which should be in
../tapes/rpf140.het
| Expected RC |
|---|
| 0 |
This job will copy the RPF load modules to SYS2.CMDLIB, which is part of LNKLST00.
| Expected RC |
|---|
| 0 |
Don't you just miss SDSF? Well, we don't have the source for it and can't build it. but there is this thing called CBT Tape, which is very useful. Among others I found the program QUEUE, aka QUE, aka (you guessed it) Q. It allows you to look at the contents of the spool, browse output datasets, Syslog, etc. It is not as comfortable as SDSF, but much better than the OUTPUT command
Just run this job. It will place the Q source into your system. You may want to invoke RPF and look into JES2.QUEUE.ASM. There are two members where you might want to do some customization: QCOMMON and QSTART. Read the comments in those members on where modifications might be required.
If you kept the names as suggested in this article you will not need any changes and can immediately compile queue by either submitting member COMPILE from the JES2.QUEUE.ASM dataset or by running job
| Expected RC |
|---|
| 0 |
This job will run a series of assemblies (35 altogether). Each and everyone should get a RC=0, if not, the jobstream aborts on the first non-zero return code. After all assemblies are done, the program is linked into SYS2.CMDLIB
| Expected RC |
|---|
| 0 |
In order to see which functions are available, press PF1 to show the first help screen. When you press the ENTER key, the next help screens will be displayed
RPF is great - but it has one drawback! It only allows editing of fixed length
datasets. Varying length records are not supported. If you need those, you might
consider installing FSE - the Full Screen Editor.
Table of Contents
Open the file fse.sed in the directory conf. It contains control statements for the sed utility. Follow the comments and instructions and change this member to fit your needs, then save it.
Now run command script ins0160. When this has finished, two new jobs will have appeared in your jcl directory:
Run this job unchanged, and it will create the FSE libraries on your system.
| Expected RC |
|---|
| 0 |
Which does a bunch of assemblies and links, all of which should be successful. Note, that an SVC will be linked into SYS1.LPALIB. This means that you cannot use FSE until you have done an IPL with the CLPA option.
| Expected RC |
|---|
| 0 |
If you do not know, what the CBT tape is - go to http://www.cbtorg.com to find out more. For MVS S3.8 the older CBT tapes are really interesting, as opposed to the newer ones, which use such new-fangled stuff as 31 bit addressing, weird set of additional registers, etc. etc. p. p.
I have created four 3350 disk volumes packed with PDS datasets containing the files from the following CBT tapes:If you haven't done so already, shut down MVS and the Hercules engine. Then run ins0170,
This installation script will ask you if you want to copy the DASD image files to your hard disk, or if you want to use the CBT volumes directly off the CD-ROM installation medium. It will also add 4 volumes to your MVS configuration file:
D U,DASD,ONLINEYou should now see the newly added disk volumes.
Then run
This job will import and connect the two CBT catalogs on the CBT volumes, and will define the three high level qualifiers used so far:
| Expected RC |
|---|
| 0 |
The file names on the CBT volumes are CBTxxx.FILEnnn, where xxx is the CBT volume (OVF, 249, 429), and nnn is the file number on that CBT tape. Now, how do you find out, which file contains what? well, here is a list:
Now the MVS System is somewhat useable. But - useable for what? Do you want
to develop applications? Well, no problem. You can use any language you like, as long as
it is assembler.... What if you are more fluent in another language? The MVS distribution
does not contain any other languages, but the OS/360 distribution does. We can salvage parts of
OS/360 and place it into MVS. This work has been done by Jay Moseley, and I have reproduced
it here for the sake of completenes. His original work can be found at his page
Compilers for MVS 3.8
Table of Contents
Running job lng0010 will load the following compilers from tape to your MVS system:
The job will also install the SORT utility. A discussion of the sort parameters etc can be found in the Sort User's Guide manual.
| Expected RC |
|---|
| 0 |
Running job lng0010 will compile and run simple "Hello World" type programs with each compiler, and will sort a dataset of a few hundred records
Shut down MVS and the hercules engine, and then run ins0180 to create the snapshot # 6. This is the last snapshot for this installation. If you are satisfied with this build of the MVS system, you can delete the earlier snapshots to reclaim some disk space.
© BSP Beratung, Schulung, Projekte GmbH