Re: Extra lines added to ONCONFIG by onparams - B169031

Malcolm Weallans <malcolm.iiug@btopenworld.com> wrote: >We are experiencing a phenomenon whereby extra lines >are being added to the ONCONFIG file. After these >lines have been added we are unable to use onparams >again or to start IDS again until we delete the extra >lines. The extra lines are copies of the 2 VPCLASS >lines that already exist in the onconfig file. > >If we resize the physical log we get one pair of lines >added. If we drop or add logical logs we get two >pairs of lines added. I have also observed that if we >edit the onconfig file using onmonitor then we get the >lines added as well. > >Environment >IDS 9.40.FC3 >IBM AIX 5.2 > >I have tested the same procedures on LINUX and that >doesn't add the extra lines. > >Investigations of cdi, IBM support pages, and IIUG IDS >forum have so far revealed that there has been one >other sighting of this phenomena, but that path has >led us nowhere. > >Any ideas? Dear Malcolm, I've created bug B169031 for this since I can reproduce the problem (extra lines in $ONCONFIG after using onparams) on Solaris 8 with IDS 9.40.UC4 when I modify the onconfig.std to include VPCLASS lines, including the 'failure to restart' part. Without the VPCLASS entries in onconfig.std, there is no problem. (Moral - maybe it isn't a good idea to hack onconfig.std after all!) Please can you send me a copy of both your $ONCONFIG file and your onconfig.std file - though what I really need is, I think, confirmation that your onconfig.std file contains at least one VPCLASS entry that is not a comment. I absolutely agree that the onconfig.std should use VPCLASS information - and only annotate in comments the older ones (B168673). I also see that the 9.40 Admin's Ref (ct1syna.pdf) doesn't list SINGLE_CPU_VP as deprecated, but it is far from obvious why you would need it if you have a VPCLASS entry such as: VPCLASS cpu,num=1,max=1 It also isn't wholly clear to me without more study why you'd use NET VPs when you get a TLI (this is Solaris) and a STR VP created - presumably to handle the TLI connections and STR connections. This is a bug in an area of code which I already have my eyes on - it's gruesome! (B169027). -- Jonathan Leffler (jleffler@us.ibm.com) STSM, Informix Database Engineering, IBM Data Management 4100 Bohannon Drive, Menlo Park, CA 94025 Tel: +1 650-926-6921 Tie-Line: 630-6921 "I don't suffer from insanity; I enjoy every minute of it!" sending to informix-list