Re: stack overflow locking sysprocplan

After more research and a number of valuable replies to my post here
and talking to a couple of Informix gurus it seems I have 4 possible
causing my sysprocplan locking problem.

1 - Cognos's Impromptu is causing the problem
2 - IBM's SDK/ODBC is causing the problem
3 - a combination of the above 2 items together are causing the
problem
4 - I have a corrupted sysmaster

We are in the process of testing a new version of the SDK (2.90 TC1) to
see if that version works better then the 2.81 TC2 and TC3 we are using
now.

We are also trying to get Cognos to look into a possible problem with
their cogdmif.ini and ifdba.dll that Savio Pinto talked about. But so
far we haven't convinced out 3rd party provider this is an issue we
need Cognos help with. (we don't have direct support to Cognos or
Informix). We are getting pretty good at playing their game of this is
consulting and not Cognos/Informix issue. So hopefully we will get to
Cognos in the next week after all the game playing.

I'm now leaning towards a corrupt sysmaster database because since
upgrading to IDS 9.4 in April oncheck -cc returns the errors below:

oncheck -cc
Sat Jul 16 00:30:00 CDT 2005
ERROR: informix.systabauth nextsize 664 != tblspace.nxtsize 128
ERROR: informix.sysconstraints nextsize 160 != tblspace.nxtsize 8
ERROR: informix.sysdefaults nextsize 128 != tblspace.nxtsize 8
ERROR: informix.sysdistrib nextsize 456 != tblspace.nxtsize 8


Since these tables are all views, and we are having sysprocplan locking
problems, it just seems to be pointing towards a corruption of
sysmaster. I wasn't around when we got IDS and created our databases
in 1998, but everyone thinks we started with IDS 7.1 and upgraded to
7.3 in 9/03 and then in April upgraded to 9.4. It is possible we
started with 7.0 then shortly afterwards upgraded to 7.1 while still in
the test and training mode, no one can remember for sure. They only
know we launched a year later on IDS 7.1.

With that all said, how do you 'rebuild' sysmaster and get rid of
the corruption? We have one instance of IDS with 5 production
databases on that instance with about a total of 2250 tables across the
5 databases.

I have found the script /informix/etc/buildsmi which looks like it
builds the sys* databases. At this point I am not sure what steps I
would take to get rid of any corruption and rebuild the sys* databases.
Can anyone enlighten me on how to do this? If I'm right about the
corruption in sysmaster this sounds like something the average DBA
would not want to tackle with out at least consulting with IBM/Informix
Support help, I'm I right on this?

John

.

Relevant Pages

  • Re: Split database
    ... In a multi-user environment, splitting is a must, in order to avoid frequent ... I've known cases of monolithic multi-user databases (mine - ... corruption incident, but even so it is always just a matter of time, and ...
    (microsoft.public.access.tablesdbdesign)
  • Re: IDE Drive failure Exch 5.5. Priv edb in a mess.
    ... I have an 'unfixed' copy of the databases, ... > Here is an article that addresses recovering from corruption. ... >> the hard disk didn't have much life left, ... >> are a large number of mails missing and many existing messages corrupt. ...
    (microsoft.public.exchange.admin)
  • Re: from Agent to Dialog to Xnews
    ... debugging software which manipulated data in Oracle and DB2 databases. ... data corruption, period. ... I used Agent for many, many years, including for binaries. ... Agent for some of the large dump groups, totally dwarfing the size of the ...
    (news.software.readers)
  • Re: flagship sunk
    ... Will never happen IDS is just another database to IBM. ... > two databases and you pick the one that best fits your needs. ... disappeared into the Purple Pile and that the brand will one day be fully ...
    (comp.databases.informix)
  • Re: Strange apparent value in decimal column using dbaccess
    ... displayed properly in dbaccess and is causing problems within our ... I've seen some vaguely similar issues before, but only on rather old versions of IDS on rather old versions of operating systems and it wasn't slashes that I saw. ... If you are able to dump the pages which contain the rows, I'd be curious to look at the hex values. ... As someone else said, you appear to have some corruption in the data, and nothing short of an update is likely to fix it. ...
    (comp.databases.informix)