Did anybody attempt to code procedures to SADUMP an lpar via OPSREXX ADDRESS HWS (BCPii)?
We can do the STOP and LOAD (IPL), but need to do the STORE STATUS somehow.
I guess it will need the HWISET at some point, but sofar have not been able the find the correct attribute?
any tips are welcome.
Marcel van Ek
As of today i have not worked before on a ticket for a client requesting information about how take a Stand-Alone Dump using the CA OPS/MVS ADDRESS HWS SENDCMD host command environment.
You can use SENDCMD to issue STOP, DEACTIVATE, ACTIVATE and LOAD (obviously not from the current CPC or LPAR).
Checked the z/OS manual to see if there are any known restrictions and the only 3 documented reasons that would prevent you from issue the Store Status function are documented in the following manual:
z/OS MVS Programming: Callable Services for High-Level Languages
Under the section "HWICMD — Issue a BCPii hardware management command" there is a "Restrictions" item that reads:
BCPii does not allow any HWICMD to be targeted to a CPC that is earlier than a z9 platform. BCPii does not allow HWICMD to be issued from within a BCPii ENF exit routine.
BCPii does not allow any HWICMD to be issued from a REXX exec running in a TSO/E or ISV-provided REXX environment.
This is also what I found in the IBM books about the procedure you should follow for taking the SAD dump:
To take a stand-alone dump:
For dumping a second level system using 3390 DASD, use the IPL command without the CLEAR operand, and do not specify LOADPARM. If using SCSI devices, first use the CP SET DUMPDEV command to define the location of the dump device using the same values as described in previous paragraph. Then use the CP IPL command with the DUMP operand.
The dumper displays messages on the system console as it progresses. For example, these messages were displayed when dumping a 10 GB second level system to a 3390 DASD at device number 555.
CP STOP CPU ALL STORE STATUSStore complete. TERM CONMODE 3215 IPL 555Booting default...Starting Stand-alone DumpDone initializing disksDumping Frame TableDumping memory:00000000 / 00000047 MB00000008 / 00000047 MB00000016 / 00000047 MB00000024 / 00000047 MB00000032 / 00000047 MB00000040 / 00000047 MB00000047 / 00000047 MBStand-alone Dump was successful.HCPGSP2629I The virtual machine is placed in CP mode due to a SIGP stop from CPU 00.
Links to the two documents where I found all these references are:
IBM Knowledge Center
Marcel, let us know if this helps or if you already tried it and encounter problems while doing it.
You couild also wait a bit to see if the rest of our community has tried something similar in the past.
Well, as I stated we've accomplished the STOP/START/DEACTIVATE/ACTIVATE/LOAD already (keeping in mind the restrictions, thus from another lpar)
It is the STORE STATUS that is more difficult, as there is no simple SENDCMD for it via HWS.
Did you complete this IBM documented step prior using the STORE STATUS command from the CA OPS/MVS ADDRESS HWS SENDCMD host command environment.
1.- Stop the system. If the system is at a disabled wait state, it is already stopped. Otherwise use the Stop All function on the HMC. For a second level system, use the CP STOP CPU ALL command.
When you do use STORE STATUS is there any response back to your program that can give us a clue? Do you want this to be handled via a Support ticket?
No, as I pointed out : I have no clue HOW to do the STORE STATUS from ADDRESS HWS....
That was my question.
ADDRESS HWS "SENDCMDCMDTEXT(„STORE STATUS‟) ENTITY(LPAR(„SAMPLE1‟)CPC(„IBM..MACHINE.SAMPLE1‟))"
Well, that isn't documented, but I tried anyway
result: (flagged some of the output as this is an open forum)
OPS0341J HWS error : ADDRESSHWS - NONZERO RETURN CODE: - RC 12 RSC: X'00B30039'
Well really annoying editor on this forum
We do not document all the commands you can send to the HMC console via the HWS interface.
That is the reason why you can not find the STORE STATUS command documented in your CA OPS/MVS manual.
Marcel, the RC=12 and reason codes are normally associated with a lack of security to execute the command.
One way to debug this error is captureing a SECTRACE you can terse and send us via a ticket.
Another option is make sure, in case you are running your code in a OSF Server, that the user running the OSF task has the authority to perform this HMC command.
There will be no violations erros in your Security product about it. This is why a SECTRACE is needed.
I understand you are not sharing some info for the messages returned since this is an open forum.
We can take this to a ticket if you feel we should at this time.
Let me know Marcel
Some additional messages appeared in OPSLOG which led me to believe STORE STATUS is not a valid command:
OPS0349T ADDRESSHWS - HiSrv SENDCMD failed. HiSrv RC: 12 ; RSC: X'560000B3'
OPS0349T ADDRESSHWS - HiSrv SENDCMD ERROR - unknown command.
Forgot to include these....
The OPS341 MSG acc to doc explains as:
RC12: A nonzero return code was received from HiSrv. The first two bytes of the reasoncode were set and an error message was issued.
x'39' last two bytes: HiSrv SENDCMD call error.
Plus, the user doing this has full authorization. Can do also IPL (LOAD).
I am reaching out internally and in your behalf CA HISRV Customer Support Level 1 to confirm your findings.
Once I know more from that group we will update this thread with an answer.
Hi Cesar. anything new to report?
No news to report as of today Marcel.
I am reaching out another team member from the Sustaining Engiineer group for HISRV.
They recently helped us with another verry similar case for advanced consultation.
As soon as I receive feedback from that team I will let you know.
Thanks for your patience Marcel
I have been in contact with the HISRV Sustining Engineer who is assisting us in this discussion.
We need more time to go over your request so we can offer you a solution.
I am expecting to post another update next week once we have a direction.
Thanks for the update Cesar.
Cesar, any news, It's been 2 months by now.....
Cesar is not available today due to illness. I believe he will be able to update you next week. Thanks for following up on this with us.
I am back in the office and I have already contacted the Sustaining Engineer form the HISRV product who is assisting us with this request.
Once we know more about your request we will let you know.
As promised I have a chance today to briefly discuss this post with our HISRV Sustaining Engineer. This is the direction I have been given as to how you should handle your original request.
Because the HMC command STORE STATUS requires additional parameters, like the dataset name where the SAD dump will be taken, what we recommend is to define a partition profile you can activate using the Address HWS interface.
You can manually activate a partition profile by using the Hardware Management Console (HMC). When you activate a partition profile, you activate a logical partition and you can also setup a SAD dump to be taken.
The system commits resources to the logical partition based on the configuration in the partition profile and starts the operating system or software that is installed on the logical partition. As part of this process you could also request a STORE STATUS to be performed as part of the profile activation. Settings in the partition profile will allow you to do this and will make easier for you to code the OPS/REXX program that uses the Address HWS interface to activate the profile. For additional information how to perform the profile activation please visit the following IBM publication:
This is a direct link to it:
Hope this gives you ideas Marcel and sorry for taking so long to respond.
Thank you Cesar, I'll discuss it internally
Sounds like a plan Marcel
Keep us posted.
The main show stopper here is how it currently works with a HWI_CMD_LOAD call.
The current HWI_CMD_LOAD design is to automatically do a CLEAR by default; if this works fine for an IPL, it is obviously not suited for IPLing SADMP.
There is a RFE opened requesting this support for stand-alone dump, titled:
'Want to start/IPL a Standalone Dump using BCPII'
I suggest you review the RFE and vote, the more votes the more weight it puts on the request."