>We are still at 1.8, but I had intended to use this functionality as soon as we >have migrated to 1.10. We are loosing SMF data regularly during shutdown of >*one* system at a time, because of all the SMF30 termination records that >get written at DB2 and IMS shutdown. We cannot increase the SMF buffers >any further and we are at the end of any tuning for the SMF data sets. We >are at the point where we stagger the DB2/IMS shutdown in hopes that the >dumping of SMF records is through when the next comes. > >I had wondered how LOGGER would handle offload of a huge amount of SMF >data right during shutdown. Do you also see this problem if you only shut down >one system? Or only when there is no system left that can do the actual >offload from the CF? > >And may I enquire why you regularly do sysplex-wide IPLs? > Back when our auditors press us to turn on UAUDIT for RACF, we found during some times when so many RACF records were being sent to SMF it could not keep up; even though the MANx data sets were empty. Thus SMF said data was lost but did not tell us whose (was RACF). Auditors were not pleased about this and therefore IBM's solution was the LOGGER. So for our journey into the SWAMP. I know my guys have created a bunch of REXX routines to cover some of the sortcomings of how the process should work.
Our present challenge in the swamp was at shutdown when "Z EOD" is entered there is no communication to the LOGGER to wait for him/her/it to complete what has to be done so when OPS re'ipls, then nothing is lost. IBM gave us a bunch of commands to have the operator enter although so far these have not worked. Plus even though the LOGGER says it is done, it is actually still off doing its thing with no indication when it honestly is completed. Level 2 agrees with us whole heartily but needed us to open a marketing request which I understand forces different parts of IBM to take notice and communicate with each other. I think this kinda assigns a referee who can get the attention of all parties on behalf of the customer. So in general, shutting down any LPAR can get you in trouble with the "hosing up" LOG files for that system. The way I am told is each system in a SYSLPLEX does its own thing with its LOG files (am by no means the expert for the techies are more into this than I). We have been having some building power upgrades and without a generator, then indeed we have to put everything to bed when it happens. Then some EMC DASD upgrades with multiple BIN file changes over a number of weekends caused some more total down time. When the swamp has been turned into something manageable, then I will give an update of how things can work. I understand the rationale to change the way SMF was offloaded which goes back my days in OS/MVT. But I think IBM was remiss in not understanding the full ramifications of how they have sites implement the LOGGER. If they would have talked to some savy users, then could have made the journey much smoother. So now, we do it the taumatic way which keeps life interesting here. jim ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
