The OP wrote "at the moment Dallas has IPCS hosed up and unusable on our machine." So the questions is what the choices are until it's fixed.
FWIW, I would certainly agree that problem analysis is much easier with IPCS, and I am normally very critical of the use of 19th Century formatted dumps. -- Shmuel (Seymour J.) Metz http://mason.gmu.edu/~smetz3 ________________________________________ From: IBM Mainframe Discussion List [[email protected]] on behalf of Mario Bezzi [[email protected]] Sent: Thursday, January 14, 2021 4:54 AM To: [email protected] Subject: Re: Why LE taking Transaction Dump rather than CEEDUMP? If you have access to IPCS, an IP VERBX LEDATA 'ALL' command will give you a formatted dump similar to the CEEDUMP you are used to. mario On 1/13/21 5:58 PM, Seymour J Metz wrote: > First, without a functioning IPCS all choices are bad. I would probably take > a SYSUDUMP or SYSABEND, look at the RB chains and look at the trace table, > but browsing that from SPOOL is a lot less convenient from IPCS and I don't > know whether your shop is including the trace table. > > Have you done anything to increase CPU or virtual storage consumption? Is > there anything unusual in the log? > > > -- > Shmuel (Seymour J.) Metz > http://mason.gmu.edu/~smetz3 > > ________________________________________ > From: IBM Mainframe Discussion List [[email protected]] on behalf of > Charles Mills [[email protected]] > Sent: Wednesday, January 13, 2021 9:59 AM > To: [email protected] > Subject: Why LE taking Transaction Dump rather than CEEDUMP? > > I have a medium-sized C++ application under development (conventional > batch). I have gotten fairly used to and fairly fond of debugging S0C4's and > the like with the LE CEEDUMP (which pinpoints that sort of error down to the > source code file and line number). > > Suddenly -- and I can't quite define "suddenly" ("I didn't change anything") > -- ABENDs have started instead producing a Transaction Dump, which I find > much less convenient, especially since at the moment Dallas has IPCS hosed > up and unusable on our machine. (If I code a SYSUDUMP DD I also get one of > those, which is even less convenient for higher level language problems.) > > Why would LE start issuing Transaction Dumps rather than CEEDUMPs? There are > no TRAP(OFF) or anything similar in the source code or in the LE runtime > options. A CEEDUMP DD statement makes no difference. There is no ESTAE > issued by my code. How the heck do I get LE back to a CEEDUMP, or what the > heck should I be looking for? > > z/OS V2R4 > > +CEE3798I ATTEMPTING TO TAKE A DUMP FOR ABEND U4087 TO DATA SET: > blah.blah.blah > IGD100I 03DE ALLOCATED TO DDNAME SYS00003 DATACLAS ( ) > > IEA822I COMPLETE TRANSACTION DUMP WRITTEN TO blah.blah.blah > > +CEE3797I LANGUAGE ENVIRONMENT HAS DYNAMICALLY CREATED A DUMP. > > +CEE0374C CONDITION=CEE3204S TOKEN=00030C84 59C3C5C5 00000000 455 > > WHILE RUNNING PROGRAM myprogrm > > AT THE TIME OF INTERRUPT > > PSW 078D1400 9AE4AC66 > > GPR 0-3 FFFFFD09 FFFFFD09 1B5D522C 1AE4ABF2 > > GPR 4-7 1AE4ACF0 1AE541C0 1AE541C0 1AE541D3 > > GPR 8-B 1B584878 00000008 1B5BD2AC 1B5E68D0 > > GPR C-F 1B5771D8 1B584A58 1B5F1FAD 1B584A28 > > > Charles > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > > ---------------------------------------------------------------------- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to [email protected] with the message: INFO IBM-MAIN > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
