Keyboard Shortcuts
Likes
- H390-MVS
- Messages
Search
Re: Difficulties with TCP/IP and data sets
IPL with the starter system.
toggle quoted message
Show quoted text
Create a new name, copy the members over. Rename old dataset. Rename new dataset. IPL with normal system. On Sat, Dec 26, 2020 at 12:36 PM Pedro Pinheiro <prppedro@...> wrote:
--
Mike A Schwab, Springfield IL USA Where do Forest Rangers go to get away from it all? |
Re: Working with ISAM Data Sets
¿ªÔÆÌåÓýStandard practice for production batch files in all the shops I have ever worked was to initialize files (ISAM or VSAM included) with a header record with a binary-zeroes key value and a date and sometimes also a time as the data. ? Random addition of data with any key value after that initial header load is then possible, though not the best practice if large amounts of additional records are needed. ? At my first MVT shop (mid-1970¡¯s), we found that loading large amounts of data to an existing (already loaded with other data) ISAM file was actually most efficiently done in DESCENDING key order, due to the way ISAM works internally. ? HTH ? Peter ? From: [email protected] <[email protected]> On Behalf Of Rupert Reynolds
Sent: Saturday, December 26, 2020 12:39 PM To: [email protected] Subject: Re: [H390-MVS] Working with ISAM Data Sets ? Standard practice when I was involved was to use REPRO to copy dummy data into any new VSAM dataset, to initialise it. ? I can't remember whether it was one record (with a low key, to be ignored or deleted later), or literally no input data at all. ? I'd agree VSAM is usually the way to go, but it still needs initialising. Well, it did when I was using BLDVRP, GENCB, MODCB macros and the like :-) ? Roops ? On Sat., Dec. 26, 2020, 16:32 Joe Monk, <joemonk64@...> wrote:
|
Re: Working with ISAM Data Sets
High values is still required for various system datasets.
On Sat, Dec 26, 2020 at 11:38 AM Rupert Reynolds <rupertreynolds@...> wrote:
-- Mike A Schwab, Springfield IL USA Where do Forest Rangers go to get away from it all? |
Re: Difficulties with TCP/IP and data sets
Thank you, both of you. It works well either way. What's more intriguing is that I didn't find it easily in the manuals and Redbooks. Anyway, there's, now, a second issue. The dataset is accessible, although I can't really write anything to it because it reached its limits (who the heck tights a configuration data set that much during the generation?). Now I know the most feasible way of doing that is just creating another dataset and moving the files over. But how am I supposed to do that with a system dataset, which is currently open by the system? About what
Rahim said: thank you, but it seems not to be the problem, at all... On Sat, Dec 26, 2020 at 2:30 PM Rupert Reynolds <rupertreynolds@...> wrote:
|
Re: Working with ISAM Data Sets
Standard practice when I was involved was to use REPRO to copy dummy data into any new VSAM dataset, to initialise it. I can't remember whether it was one record (with a low key, to be ignored or deleted later), or literally no input data at all. I'd agree VSAM is usually the way to go, but it still needs initialising. Well, it did when I was using BLDVRP, GENCB, MODCB macros and the like :-) Roops On Sat., Dec. 26, 2020, 16:32 Joe Monk, <joemonk64@...> wrote:
|
Re: Difficulties with TCP/IP and data sets
Good point, but I prefer to leavw the prefix, and simply use apostrophes around dataset names that don't start with it. sub d(myjob) is easier to type than sub roopy.d.cntl(myjob) Same with? ex d(myclist) compared with ex roopy.d.clistl(myclist) call d(myprog) and test d(myprpg) likewise :-) Roops On Sat., Dec. 26, 2020, 16:33 Joe Monk, <joemonk64@...> wrote:
|
Re: Difficulties with TCP/IP and data sets
TCPIP is not active by default in Hercules.? The following may give you some clues about how to activate it.? The availability and flavor of the instruction is controlled by three elements: 1.? The compile time FEATURE_TCPIP_EXTENSION definition: To allow enabling of the TCPIP instruction, the featxxx.h header files corresponding to the relevant architectures (feat370.h, feat390.h, feat900.h) must contain the statement: ``` ??????????????? #define? FEATURE_TCPIP_EXTENSION ``` As shipped, SDL Hyperion 4.2 has this feature defined _only_ for the S/370 architecture (feat370.h). 2.? By default, the TCPIP instruction is disabled at run time, regardless of whether the compile time FEATURE_TCPIP_EXTENSION was defined or not.? If FEATURE_TCPIP_EXTENSION was defined for the architecture currently being used, the supervisor state flavor of the TCPIP instruction can only be enabled by issuing the command: ``` ??????????????? facility enable HERC_TCPIP_EXTENSION ``` before IPLing the guest operating system. 3.? If the problem state variation of the TCPIP instruction is desired, then after having enabled the supervisor state flavor as described above, the command: ``` ??????????????? facility enable HERC_TCPIP_PROB_STATE ``` must then _also_ be issued before IPLing the operating system. regards; Rahim ?? ??
On Saturday, December 26, 2020, 10:27:25 AM CST, Pedro Pinheiro <prppedro@...> wrote:
Hi, all I have a rather dumb question, I recognize. But I've been struggling with it for the past day. There's a data set (CENTER.PARMLIB) in which TCPPROF member is contained. TSO MVSCUST opens this file for me. And I can also access it through DSLIST. Though I can't access it anywhere. Any other ISPF utilities just tells me either the ¡°Data set is not cataloged¡± or (if I specify the volume) ¡°Data set is not found¡±. That's quite problematic because, on top of all that, system have been abending on me, crying 'bout lack of space in the dataset. But I can't edit CENTER.PARMLIB allocation because all the tools which seems to deal with that can't find the data set. Also, when using the ¡°ALLOC¡± command, it infers the dataset is prefixed the current login, SYSPROG1. So, is there any way out of this mess? Thanks, Att., -trp |
Re: Difficulties with TCP/IP and data sets
TSO PROFILE NOPREFIX. Then try to access ?your dataset. Joe On Sat, Dec 26, 2020 at 10:27 AM Pedro Pinheiro <prppedro@...> wrote:
|
Re: Working with ISAM Data Sets
Yeah ... DONT use ISAM. You will regret?it if you do so. USE VSAM KSDS instead. Easier and less complicated. If you need some pointers let me know. Joe On Sat, Dec 26, 2020 at 9:59 AM Patrik Schindler <poc@...> wrote: Hello, |
Difficulties with TCP/IP and data sets
Hi, all I have a rather dumb question, I recognize. But I've been struggling with it for the past day. There's a data set (CENTER.PARMLIB) in which TCPPROF member is contained. TSO MVSCUST opens this file for me. And I can also access it through DSLIST. Though I can't access it anywhere. Any other ISPF utilities just tells me either the ¡°Data set is not cataloged¡± or (if I specify the volume) ¡°Data set is not found¡±. That's quite problematic because, on top of all that, system have been abending on me, crying 'bout lack of space in the dataset. But I can't edit CENTER.PARMLIB allocation because all the tools which seems to deal with that can't find the data set. Also, when using the ¡°ALLOC¡± command, it infers the dataset is prefixed the current login, SYSPROG1. So, is there any way out of this mess? Thanks, Att., -trp |
Working with ISAM Data Sets
Hello,
"The records that comprise a newly created data set must be presented for writing in ascending order by key." Source: Does this mean that I need to at least write one record with some arbitrary key value, and later (DISP=MOD) can add more records which are not necessarily in order with the key value already in the first record? I want to find some practical use for a batch workflow involving collection of data and generating reports of it in TK4-. I decided to postpone VSAM to later when I feel skilled enough to create interactive applications using KICKS. Thanks! :wq! PoC |
Re: Show disk usage in TK4-
When Hercules writes the data from memory cache to disk, it compresses
toggle quoted message
Show quoted text
the data and writes at a new location. The old location is marked as available and the files is eventually compressed to remove unused space resulting in a smaller file on the host. An delete on the host just marks the tracks as available in the VTOC (or free space map in the VTOCINDEX). An erase before delete would do both (I guess). Mainframe programs that are I/O limited on actual hardware are CPU limited on Hercules, since it has so much disk cache in memory. Early versions of Hercules did cause mainframe programs problems due to I/O completing so fast. On Mon, Dec 21, 2020 at 6:38 AM Patrik Schindler <poc@...> wrote:
--
Mike A Schwab, Springfield IL USA Where do Forest Rangers go to get away from it all? |
Re: Show disk usage in TK4-
Hello Joe,
Am 21.12.2020 um 14:15 schrieb Joe Monk <joemonk64@...>: "The space recovery routine relocates track or block group images towards the beginning of the file, causing free space to move towards the end of the file. When a free space reaches the end of the file, it 'falls off', that is, the file size is reduced."Thanks a lot, now it¡¯s clear to me. :wq! PoC |
Re: Show disk usage in TK4-
Hercules uses PC based disk files. The format inside those PC based disk files is mainframe CKD or FBA. The "host file" that Mike was referring to is the PC based disk file. Since there is no need to statically allocate the entirety?of the DASD space (called "thin provisioning" in SAN land), the hercules disk process can dynamically allocate and release space from its emulated mainframe DASD files without issue... "The space recovery routine relocates track or block group images towards the beginning of the file, causing free space to move towards the end of the file. When a free space reaches the end of the file, it 'falls off', that is, the file size is reduced." Joe On Mon, Dec 21, 2020 at 6:38 AM Patrik Schindler <poc@...> wrote: Hello Mike, |
Re: Show disk usage in TK4-
Hello Mike,
Am 20.12.2020 um 18:57 schrieb Mike Schwab <Mike.A.Schwab@...>: Hercules CCKD(64) releases the space to condense the host file in the background.May I ask what exactly you mean by ?release¡°? Overwrite them with a constant value, so a high compression ratio is achieved? On I/O-heavy loads, could it be beneficial to not use compression at all? Moshix (the mainframe video guy) claimed, that today, CPUs are so fast, that it barely matters. What¡¯s your opinion? :wq! PoC |
Re: Show disk usage in TK4-
Jeffrey Melton wrote:
But regardless of whether you actually have physical[...] Just making sure I understand the ramifications of allPrecisely! You understand the situation perfectly. Using an inefficient block size will waste space on the volume, causing it to hold less data than it otherwise could. The operating system would not be able to fit as much data on the volume (dasd) than it otherwise could have fit had a more efficient block size been chosen, and this is *regardless* of whether the dasd is a Hercules uncompressed CKD dasd or a compressed CCKD/CCKD64 dasd, because Hercules DOES strictly enforce an emulated dasd device type's maximum track size limitation. -- "Fish" (David B. Trout) Software Development Laboratories mail: fish@... |
Re: Show disk usage in TK4-
¿ªÔÆÌåÓýExactly correct.? You¡¯ve got it. ? Peter ? From: [email protected] <[email protected]> On Behalf Of Jeffrey Melton via groups.io
Sent: Monday, December 21, 2020 1:20 AM To: [email protected] Subject: Re: [H390-MVS] Show disk usage in TK4- ? But regardless of whether you actually have physical "wasted" space on your real disk drive by using compressed virtual devices, if you choose a suboptimal block size, you still won't be making the most of the (even compressed) virtual devices you do allocate in your IOGEN, correct?? Granted there is a tradeoff, but say you really screw up and set up a terrible block size for your files that causes MVS to "waste" half of the "available" space of a given virtual DASD.? Compressed or non-compressed, that would still mean that MVS thinks that virtual device is "full" when it really isn't, it's just that the existing files are allocated terribly out of whack, and you'll end up having to create and maintain more virtual "devices" in your emulated mainframe's IOGEN than would normally be necessary, correct? --._,_._,_ |
Re: Show disk usage in TK4-
But regardless of whether you actually have physical "wasted" space on your real disk drive by using compressed virtual devices, if you choose a suboptimal block size, you still won't be making the most of the (even compressed) virtual devices you do allocate in your IOGEN, correct?? Granted there is a tradeoff, but say you really screw up and set up a terrible block size for your files that causes MVS to "waste" half of the "available" space of a given virtual DASD.? Compressed or non-compressed, that would still mean that MVS thinks that virtual device is "full" when it really isn't, it's just that the existing files are allocated terribly out of whack, and you'll end up having to create and maintain more virtual "devices" in your emulated mainframe's IOGEN than would normally be necessary, correct?
Just making sure I understand the ramifications of all this correctly, Jeff.? |
Re: Show disk usage in TK4-
Normally, the track is not erased on delete. It is a rarely used
toggle quoted message
Show quoted text
option on VSAM files because it can take a while. There are some programs that can go through and erase empty tracks later. Hercules CCKD(64) releases the space to condense the host file in the background. On Sun, Dec 20, 2020 at 10:54 AM Patrik Schindler <poc@...> wrote:
--
Mike A Schwab, Springfield IL USA Where do Forest Rangers go to get away from it all? |