Re: Informix v7.24 on HP-UX 10.20
Eric v7.24 uc6. Phew! I don't get a very warm feeling from Informix Tech Support - the whole thing seems full of holes on HP-UX. I previously worked at a Solaris site and I have to say that, both generally and specific to Informix, Solaris seems streets ahead. The bug fix I quoted is dated 31/3/98. A pity Tech Support took 8 days to mention it from the time that I first reported technical problems. Once again, at my previous site we seemed to get better support from Informix. Could it be anything to do with the fact that then I was at the world-renowned Chase Manhattan and now I work at a small-to-middling size grocery chain? Neil -----Original Message----- From: Eric_Melillo@agsea.com <Eric_Melillo@agsea.com> To: Neil Truby <ntruby@netcomuk.co.uk> Cc: informix-list@iiug.org <informix-list@iiug.org> Date: 20 January 1999 14:53 Subject: Re: Informix v7.24 on HP-UX 10.20 > >Hey Neil -- > >Eric here... Please tell me that you are either using 7.24.UC4 or greater, > OR that you have NOAGE=0. > >NOTE: THIS ONLY APPLIES TO HP ( AFAIK ) > >The reason is: Bug / Defects 55545 and 68612. Which read: > Setting the NOAGE onconfig paramater equal to 1 causes the CPU VP > to take all available cpu resources when AIO VPs are introduced to > REAL-TIME class. > >============== >My experience is that it does all of that and then PANIC reboots the >system. >============== > >The work around is to manually reschedule the VPs (all but AIO's). I have >the process, if you need it. > >The fix for the problem is in the following minimum releases: 7.23.UC9, >7.24.UC4, and 7.30.UC2. > >The KAIO bug you refer to ( 92132 ) is fixed in the following releases: >Patched 7.24.UC5XQ, Patched 7.24.UC6XB, and the scheduled GA >of 7.24.UC8. > >Yes, I experience the effects of this bug also. It is nice to know that I >have >a kindred soul that I can commiserate with ( the same bugs, LAWSON, etc ). > >"Talk" to you later. > >Eric P. Melillo >Sr. Database Administrator >Associated Grocers, Inc. >Seattle, WA 98118 >vmail: (206) 767-8178 >email: eric_melillo@agsea.com > > > > > > >"Neil Truby" <ntruby@netcomuk.co.uk> on 01/20/99 01:32:18 AM > >Please respond to "Neil Truby" <ntruby@netcomuk.co.uk> > >To: informix-list@iiug.org >cc: (bcc: Eric Melillo/AGInc) >Subject: Informix v7.24 on HP-UX 10.20 > > > > >We've recently upgraded to this version and had a lot of problems with it. >I've posted pleas for help on some of these problems. We seem to have got >to the bottom of them now, so may I share my experiences? I would point >out >that, as far as I know, these problems are specific to HP-UX: > >Originally we found that IDS would hang after a few minutes' operation. >This problem was eventually remedied (after five weeks) by HP, who >suggested, inter alia, setting RESIDENT=1. We've had no further hanging >problems. > >Once we did go live we found that we had greatly reduced performance. This >was most noticeable (in reality probably most measurable) in our regular >batch suites and. for some reason, level 1 archives, whose elapsed times >increased by a factor of three or more. Informix have now alerted me to >bug >warning 92132, which warns of severe I/O performance degradation using kaio >on some versions of v7.2x. I have disabled kaio on our production engine >and, as if by magic, the problems have gone away. > >So be warned! > >Neil Truby >aracnet Limited >Weybridge, UK > > > > > > > > > > >
In article <785bo9$32b$1@news.xmission.com>, Neil Truby <ntruby@netcomuk.co.uk> writes > >Eric > >v7.24 uc6. Phew! > >I don't get a very warm feeling from Informix Tech Support - the whole thing >seems full of holes on HP-UX. I previously worked at a Solaris site and I >have to say that, both generally and specific to Informix, Solaris seems >streets ahead. > >The bug fix I quoted is dated 31/3/98. A pity Tech Support took 8 days to >mention it from the time that I first reported technical problems. Once >again, at my previous site we seemed to get better support from Informix. >Could it be anything to do with the fact that then I was at the >world-renowned Chase Manhattan and now I work at a small-to-middling size >grocery chain? > >Neil > I keep on saying go to 7.30.UC5-1!!! PS Someone mentioned 7.30.UC6 is that generally available yet??? PPS I saw a bug fix in Techinfo that said it was fixed in 7.30.UC7! Is that available yet?? >-----Original Message----- >From: Eric_Melillo@agsea.com <Eric_Melillo@agsea.com> >To: Neil Truby <ntruby@netcomuk.co.uk> >Cc: informix-list@iiug.org <informix-list@iiug.org> >Date: 20 January 1999 14:53 >Subject: Re: Informix v7.24 on HP-UX 10.20 > > >> >>Hey Neil -- >> >>Eric here... Please tell me that you are either using 7.24.UC4 or greater, >> OR that you have NOAGE=0. >> >>NOTE: THIS ONLY APPLIES TO HP ( AFAIK ) >> >>The reason is: Bug / Defects 55545 and 68612. Which read: >> Setting the NOAGE onconfig paramater equal to 1 causes the CPU VP >> to take all available cpu resources when AIO VPs are introduced to >> REAL-TIME class. >> >>============== >>My experience is that it does all of that and then PANIC reboots the >>system. >>============== >> >>The work around is to manually reschedule the VPs (all but AIO's). I have >>the process, if you need it. >> >>The fix for the problem is in the following minimum releases: 7.23.UC9, >>7.24.UC4, and 7.30.UC2. >> >>The KAIO bug you refer to ( 92132 ) is fixed in the following releases: >>Patched 7.24.UC5XQ, Patched 7.24.UC6XB, and the scheduled GA >>of 7.24.UC8. >> >>Yes, I experience the effects of this bug also. It is nice to know that I >>have >>a kindred soul that I can commiserate with ( the same bugs, LAWSON, etc ). >> >>"Talk" to you later. >> >>Eric P. Melillo >>Sr. Database Administrator >>Associated Grocers, Inc. >>Seattle, WA 98118 >>vmail: (206) 767-8178 >>email: eric_melillo@agsea.com >> >> >> >> >> >> >>"Neil Truby" <ntruby@netcomuk.co.uk> on 01/20/99 01:32:18 AM >> >>Please respond to "Neil Truby" <ntruby@netcomuk.co.uk> >> >>To: informix-list@iiug.org >>cc: (bcc: Eric Melillo/AGInc) >>Subject: Informix v7.24 on HP-UX 10.20 >> >> >> >> >>We've recently upgraded to this version and had a lot of problems with it. >>I've posted pleas for help on some of these problems. We seem to have got >>to the bottom of them now, so may I share my experiences? I would point >>out >>that, as far as I know, these problems are specific to HP-UX: >> >>Originally we found that IDS would hang after a few minutes' operation. >>This problem was eventually remedied (after five weeks) by HP, who >>suggested, inter alia, setting RESIDENT=1. We've had no further hanging >>problems. >> >>Once we did go live we found that we had greatly reduced performance. This >>was most noticeable (in reality probably most measurable) in our regular >>batch suites and. for some reason, level 1 archives, whose elapsed times >>increased by a factor of three or more. Informix have now alerted me to >>bug >>warning 92132, which warns of severe I/O performance degradation using kaio >>on some versions of v7.2x. I have disabled kaio on our production engine >>and, as if by magic, the problems have gone away. >> >>So be warned! >> >>Neil Truby >>aracnet Limited >>Weybridge, UK >> >> >> >> >> >> >> >> >> >> >> > -- David Williams
Neil Truby wrote: > The bug fix I quoted is dated 31/3/98. A pity Tech Support took 8 days to > mention it from the time that I first reported technical problems. Once > again, at my previous site we seemed to get better support from Informix. > Could it be anything to do with the fact that then I was at the > world-renowned Chase Manhattan and now I work at a small-to-middling size > grocery chain? No, I feel pretty safe in saying that it doesn't matter who your company is, with a *possible* (but by no means definite) exception if you are a Regency or Enterprise Support customer. I'd say it's most closely related to who (which engineer) picks up your call, and not who you are. June -- june_t@hotmail.com Grounded in Palo Alto, living on M&M's (plain)