This is a discussion on 10.0 xC6 within the Informix forums, part of the Database Server Software category; --> Does anyone have any experiences - good or bad - on IDS 10.0xC6? We have several customers on FC5 ...
| |||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| Does anyone have any experiences - good or bad - on IDS 10.0xC6? We have several customers on FC5 and all seems fine so far. We have one moving from 32-bit (10.0UC4W3) to 64-bit and I'm trying to decide if to move them to FC5, or give FC6 a go. All opinion gratefully received! -- Neil Truby t:01932 724027 Director m:07798 811708 Ardenta Limited e:neil.truby@ardenta.com |
| |||
| On Feb 4, 5:57 pm, "Neil Truby" <neil.tr...@ardenta.com> wrote: > Does anyone have any experiences - good or bad - on IDS 10.0xC6? > > We have several customers on FC5 and all seems fine so far. We have one > moving from 32-bit (10.0UC4W3) to 64-bit and I'm trying to decide if to move > them to FC5, or give FC6 a go. > > All opinion gratefully received! > > -- > Neil Truby t:01932 724027 > Director m:07798 811708 > Ardenta Limited e:neil.tr...@ardenta.com I have had no problems with 10.FC6, I am almost done upgrading all servers to this release except peoplesoft of course. I had no issues with onmonitor either. As for 64-bit I moved to it as soon as it was available and never looked back. I'm on HP/UX. On another note you will need this release for DST (march 11) unless you are running an OS platforn that supoorts using the utility. Good-luck, DBAGIRL |
| |||
| "dbagirltx" <dbagirltx@yahoo.com> wrote in message news:1170789326.225772.26580@a75g2000cwd.googlegro ups.com... > On Feb 4, 5:57 pm, "Neil Truby" <neil.tr...@ardenta.com> wrote: >> Does anyone have any experiences - good or bad - on IDS 10.0xC6? >> >> We have several customers on FC5 and all seems fine so far. We have one >> moving from 32-bit (10.0UC4W3) to 64-bit and I'm trying to decide if to >> move >> them to FC5, or give FC6 a go. >> >> All opinion gratefully received! >> >> -- >> Neil Truby t:01932 724027 >> Director m:07798 811708 >> Ardenta Limited e:neil.tr...@ardenta.com > > I have had no problems with 10.FC6, I am almost done upgrading all > servers to this release except peoplesoft of course. I had no issues > with onmonitor either. As for 64-bit I moved to it as soon as it was > available and never looked back. I'm on HP/UX. On another note you > will need this release for DST (march 11) unless you are running an OS > platforn that supoorts using the utility. > > Good-luck, DBAGIRL Thanks for that. What is DST (march 11) please? |
| |||
| Neil Truby wrote: > "dbagirltx" <dbagirltx@yahoo.com> wrote in message > news:1170789326.225772.26580@a75g2000cwd.googlegro ups.com... >> On Feb 4, 5:57 pm, "Neil Truby" <neil.tr...@ardenta.com> wrote: >>> Does anyone have any experiences - good or bad - on IDS 10.0xC6? >>> >>> We have several customers on FC5 and all seems fine so far. We have one >>> moving from 32-bit (10.0UC4W3) to 64-bit and I'm trying to decide if to >>> move >>> them to FC5, or give FC6 a go. >>> >>> All opinion gratefully received! >>> >>> -- >>> Neil Truby t:01932 724027 >>> Director m:07798 811708 >>> Ardenta Limited e:neil.tr...@ardenta.com >> I have had no problems with 10.FC6, I am almost done upgrading all >> servers to this release except peoplesoft of course. I had no issues >> with onmonitor either. As for 64-bit I moved to it as soon as it was >> available and never looked back. I'm on HP/UX. On another note you >> will need this release for DST (march 11) unless you are running an OS >> platforn that supoorts using the utility. >> >> Good-luck, DBAGIRL > > Thanks for that. > What is DST (march 11) please? > > The change in Daylight Savings Time... AFAIK north America will change the day when the DST will become active. Please check: http://www-1.ibm.com/support/docview...id=swg21248851 and Guy Bowerman blog: http://www-03.ibm.com/developerworks..._time_sho uld Regards. -- Fernando Nunes Portugal http://informix-technology.blogspot.com My email works... but I don't check it frequently... |
| |||
| "Fernando Nunes" <spam@domus.online.pt> wrote in message news:45c905d3$0$1723$a729d347@news.telepac.pt... > Neil Truby wrote: >> "dbagirltx" <dbagirltx@yahoo.com> wrote in message >> news:1170789326.225772.26580@a75g2000cwd.googlegro ups.com... >>> On Feb 4, 5:57 pm, "Neil Truby" <neil.tr...@ardenta.com> wrote: > The change in Daylight Savings Time... AFAIK north America will change the > day when the DST will become active. Ah, that's "international". I don't care! ;-) |
| |||
| On 6 Feb, 23:36, "Neil Truby" <neil.tr...@ardenta.com> wrote: > "Fernando Nunes" <s...@domus.online.pt> wrote in message > > news:45c905d3$0$1723$a729d347@news.telepac.pt... > > > Neil Truby wrote: > >> "dbagirltx" <dbagir...@yahoo.com> wrote in message > >>news:1170789326.225772.26580@a75g2000cwd.googleg roups.com... > >>> On Feb 4, 5:57 pm, "Neil Truby" <neil.tr...@ardenta.com> wrote: > > The change in Daylight Savings Time... AFAIK north America will change the > > day when the DST will become active. > > Ah, that's "international". I don't care! ;-) Well we had one crash on FC4 (The first in months and months of v10 use!) due to the bug where the engine crashes when RESIDENT+MESSAGE+SHMVIRT> SHMTOTAL that is only fixed in FC5 so....we will be getting and starting testing FC6 ASAP! I am interested in the new FASTPOLL feature in FC5 and above and the new BTSCANNER "alice" (adaptive linear index cleaning) option ! Neil, It would be nice if you could extend your BTSCANNER white paper to include "alice" scanning!! David. |
| |||
| > Neil, It would be nice if you could extend your BTSCANNER white paper > to include "alice" scanning!! Andy Watson, who writes all this stuff, is sulking because he says no-one ever responds to his cdi posts. I'll try and coax him by saying how much you liked it! |
| |||
| On 14 Feb, 03:29, "Alexey Sonkin" <alex...@cidc.com> wrote: > Hi, Neil, > > I started testing uc6/fc6 a few days ago, and ran > into a really bad problem: very big memory leak > in virtual memory segments. > > The strange thing is that 'onstat -g seg' shows, > that 800 MB is used in 6 virtual segments: > > onstat -g seg > > IBM Informix Dynamic Server Version 10.00.UC6 -- On-Line (Prim) -- > Up 1 days 12:05:35 -- 1672528 Kbytes > > Segment Summary: > id key addr size ovhd class blkused blkfree > 88997892 1381451777 44000000 791068672 241036 R 193129 3 > 89030661 1381451778 7326c000 409600000 13152 V 99992 8 > 89063430 1381451779 8b90c000 102400000 3776 V 24999 1 > 89096199 1381451780 91ab4000 102400000 3776 V 24997 3 > 89128968 1381451781 97c5c000 102400000 3776 V 25000 0 > 89161737 1381451782 9de04000 102400000 3776 V 24999 1 > 89194506 1381451783 a3fac000 102400000 3776 V 2498 22502 > Total: - - 1712668672 - - 395614 22518 > > 'onstat -g mem' only shows about 100MB in total allocated from virtual > segments. Completely unclear, what happened to other 700MB... > What does onstat -g afr say? Please let us know what Tech Support say! |
| ||||
| On 17 Feb, 04:53, "Alexey Sonkin" <alex...@cidc.com> wrote: > Hi, everybody, > > The mystery of enormous memory leak in 10.0uc6 is finally solved. > Many thanks to IBM Informix development team, namely John Lengyel, > for informal, quick and very professional investigation. > > It turned out, the problem was introduced by the new functionality: > 'Private Memory Caches for Virtual Processors'. > This functionality is targeted for Cheetah, but is introduced > in 10.0xC5. We didn't run into it with xC5, because we found another > problem in xC5 and stopped testing it. > By mistake, the new functionality is turned 'on' by default. Was this parameter not in your onconfig file? Is the problem is that the 'default' value used when there is no value in the onconfig file is too high? Since IDS allocates memory in blocks of 8k then surely setting this to something like 128k would help spinlock contention? > This functionality allowes IDS to allocate large blocks of memory > to private VP caches... > > To disable this functionality, one can set the new > $ONCONFIG parameter VP_MEMORY_CACHE_KB to 0 and restart IDS > or do it online with 'onmode -wf VP_MEMORY_CACHE_KB=0' > > Most probably, IBM will fix this very shortly. > > To reproduce the problem, one can do the following: > ----------- > -- CREATE ANSI-MODE DATABASE and connect----- > > set isolation to dirty read; > set lock mode to wait 5; > > create function str2list( > str lvarchar,separator char default ',') > returning list(lvarchar not null); > define lst list(lvarchar not null); > define dummy lvarchar; > if separator = ',' then > let dummy = 'list{'||str||'}'; > else > let dummy = 'list{'||replace(str,separator,',')||'}'; > end if; > let lst = dummy; > return lst; > end function; > commit; > > create procedure test_spl(); > define i int; > define lst list(lvarchar not null); > for i in (1 to 1000000) > let lst = str2list('1,2,3,4,5,6,7,8,9'); > end for; > end procedure; > commit; > > execute procedure test_spl(); > commit; > ------------------------------------------ > > This block would cause a 200-MB memory leak... > Actually, the problem might also exist in 10.0xC5 > > Alexey Sonkin > > > > -----Original Message----- > From: informix-list-boun...@iiug.org > > [mailto:informix-list-boun...@iiug.org] On Behalf Of d...@smooth1.co.uk > > Please let us know what Tech Support say!- Hide quoted text - > > - Show quoted text - |