Re: zFS SMF problems
On 1/06/2021 1:56 am, Colin Paice wrote: 1. The SMF 92-59 records say there are 14 sections - but there is data for only 13 sections. The field is SMF92DON-1 I had a look at my data and I can see the same thing in my 92-59 records. If there are 14 sections, the 14th section is zeros apart from the timestamp. I only see problems in records with 14 sections. I also have records with 12 or 13 sections, they don't have the problem. I don't see a partially blank name, the entire section looks like it hasn't been filled in. (The names are at the end of each section, so might look like it belongs with the next empty section if you are eyeballing a dump.) 2. In the SMF 92-11 it reports data on the individual files used. It reports the file as used, for example ./write.c I expected it to say /u/adcd/write.c. The data as is, is pretty useless, as I do not know which write.c was used. Other records have /usr/lib/nls/msg/C/fsumrcat.cat as expected (because that is what the program used) This is unfortunate, but I think it reflects the information available at the point the record is written. Given filesystem links etc. it may not even be possible to be sure of the path used to open a file if a relative path is used. The inode number and device number is probably the true identifier of a file. You can probably make reasonable guesses about a relative path by looking at directory read records from the same task. -- Andrew Rowley Black Hill Software -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: zFS SMF problems
Do you have MXG? Or have you joined Dr. Merrill's List? www.mxg.com That would be a good place to report an issue with SMF Data Or see if Cheryl Watson has seen this issue. Lizette -Original Message- From: IBM Mainframe Discussion List On Behalf Of Colin Paice Sent: Monday, May 31, 2021 8:56 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: zFS SMF problems I've been looking into the SMF records produced by zFS and have found a couple of problems. Who can I report them to? I cannot raise a PMR as I only have a zPDT license. Example problems 1. The SMF 92-59 records say there are 14 sections - but there is data for only 13 sections. The field is SMF92DON-1 2. In the SMF 92-11 it reports data on the individual files used. It reports the file as used, for example ./write.c I expected it to say /u/adcd/write.c. The data as is, is pretty useless, as I do not know which write.c was used. Other records have /usr/lib/nls/msg/C/fsumrcat.cat as expected (because that is what the program used) Colin -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: zFS SMF problems
Charles, I'm very familiar with SMF records ... I have a program which formats many of them (including SMF 92's) the 14th section has a partly blank name, and other value are all hex 00's I am planning on blogging about ZFS, concepts and how to look at the performance data; and documenting it warts and all, it would have been nice to pass it by IBM first for comment and to be able to say "fixes are coming". Colin On Mon, 31 May 2021 at 17:42, Charles Mills wrote: > Good luck! > > I share your pain on PMRs. The same is true for Dallas "Innovation Center" > customers. > > On your specific issues I would *guess* > > 1. Is it perhaps a misunderstanding on your part? Do they perhaps mean > "there are 14 possible sections; look at each relevant triplet to see if > section X is part of a given record"? Many SMF records -- SMF 30 comes to > mind -- have lots of potential sections and most records as cut do not have > them all. > > 2. I suspect Working as Designed. It is what the program opened, not the > true full path name. I suspect you are looking at a request for > enhancement, not a PMR. I hear you -- "./somefile" is pretty useless for > auditing or troubleshooting purposes -- but perhaps it is what it is. > > Charles > > > -Original Message- > From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On > Behalf Of Colin Paice > Sent: Monday, May 31, 2021 8:56 AM > To: IBM-MAIN@LISTSERV.UA.EDU > Subject: zFS SMF problems > > I've been looking into the SMF records produced by zFS and have found a > couple of problems. Who can I report them to? I cannot raise a PMR as I > only have a zPDT license. > Example problems > >1. The SMF 92-59 records say there are 14 sections - but there is data >for only 13 sections. The field is SMF92DON-1 >2. In the SMF 92-11 it reports data on the individual files used. It >reports the file as used, for example ./write.c I expected it to say >/u/adcd/write.c. The data as is, is pretty useless, as I do not know > which >write.c was used. Other records have /usr/lib/nls/msg/C/fsumrcat.cat as >expected (because that is what the program used) > > Colin > > -- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > > -- > For IBM-MAIN subscribe / signoff / archive access instructions, > send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN > -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
Re: zFS SMF problems
Good luck! I share your pain on PMRs. The same is true for Dallas "Innovation Center" customers. On your specific issues I would *guess* 1. Is it perhaps a misunderstanding on your part? Do they perhaps mean "there are 14 possible sections; look at each relevant triplet to see if section X is part of a given record"? Many SMF records -- SMF 30 comes to mind -- have lots of potential sections and most records as cut do not have them all. 2. I suspect Working as Designed. It is what the program opened, not the true full path name. I suspect you are looking at a request for enhancement, not a PMR. I hear you -- "./somefile" is pretty useless for auditing or troubleshooting purposes -- but perhaps it is what it is. Charles -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@LISTSERV.UA.EDU] On Behalf Of Colin Paice Sent: Monday, May 31, 2021 8:56 AM To: IBM-MAIN@LISTSERV.UA.EDU Subject: zFS SMF problems I've been looking into the SMF records produced by zFS and have found a couple of problems. Who can I report them to? I cannot raise a PMR as I only have a zPDT license. Example problems 1. The SMF 92-59 records say there are 14 sections - but there is data for only 13 sections. The field is SMF92DON-1 2. In the SMF 92-11 it reports data on the individual files used. It reports the file as used, for example ./write.c I expected it to say /u/adcd/write.c. The data as is, is pretty useless, as I do not know which write.c was used. Other records have /usr/lib/nls/msg/C/fsumrcat.cat as expected (because that is what the program used) Colin -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
zFS SMF problems
I've been looking into the SMF records produced by zFS and have found a couple of problems. Who can I report them to? I cannot raise a PMR as I only have a zPDT license. Example problems 1. The SMF 92-59 records say there are 14 sections - but there is data for only 13 sections. The field is SMF92DON-1 2. In the SMF 92-11 it reports data on the individual files used. It reports the file as used, for example ./write.c I expected it to say /u/adcd/write.c. The data as is, is pretty useless, as I do not know which write.c was used. Other records have /usr/lib/nls/msg/C/fsumrcat.cat as expected (because that is what the program used) Colin -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
