Cobol - Micro-Focus: bloated index

Author

Micro-Focus: bloated index

Michael Russell

2005-12-14, 6:55 pm

While I'm here ... this is unprepared, but:

Does anyone have a quick response available for an idx that suddenly
bloats to 750+Mbytes?
(and becomes unusable - can't remember the error offhand, but maybe
'badly formatted index')
There are about 120k data records; 1 primary, 2 secondaries with
duplicates allowed; manual, multiple locks allowed.

Knocking up a test program, to try to replicate, we've:
1. added 3 duplicates to each record's secondaries;
2. read, deleted and written back all data records;
3. run out of ideas

The only thing we haven't attempted is volume of transactions per second
- about
6 instances of the real program will get fired off and each will then
process up to about 1 transaction per second -
adding records, while something else gets to use & delete them, keeping
a more-or-less steady population.

The file gets reorganised once per w

. The processes run almost 15x7.

Environment - SCO unix.

Any memories stirred?

Michael

Richard

2005-12-14, 6:55 pm

> Environment - SCO unix.

I do have a client that had problems with MF Cobol on SCO unix
(Openserver I think) related to index files, and they couldn't
reproduce it in testing either. They switched to Red Hat RHEL 3 and
have no problem.

Scot Nielsen

2005-12-15, 6:55 pm

Hello Michael,

Can you provide me with some more information about your enivronment, in
particular the version of Server Express or OCDS you're using?

Also, how frequently does this problem occur and could you send me the files
on DVD?

Depending on what information you can supply and whether you can get us the
files, I'd be happy to take a closer look at this for you.

Please reply directly to my email address.

Regards.

Scot Nielsen Core Technology | File Handler and Fileshare
Micro Focus, 22-30 Old Bath Rd, Newbury, BERKS. RG14 1QN
email scot.nielsen@microfocus.com

"Michael Russell" <Michael.Russell@msn.com> wrote in message
news:43a09b68$0$17550$ed9e5944@text-readers.news.pipex.net...
> While I'm here ... this is unprepared, but:
>
> Does anyone have a quick response available for an idx that suddenly
> bloats to 750+Mbytes?
> (and becomes unusable - can't remember the error offhand, but maybe 'badly
> formatted index')
> There are about 120k data records; 1 primary, 2 secondaries with
> duplicates allowed; manual, multiple locks allowed.
>
> Knocking up a test program, to try to replicate, we've:
> 1. added 3 duplicates to each record's secondaries;
> 2. read, deleted and written back all data records;
> 3. run out of ideas
>
> The only thing we haven't attempted is volume of transactions per second -
> about
> 6 instances of the real program will get fired off and each will then
> process up to about 1 transaction per second -
> adding records, while something else gets to use & delete them, keeping a
> more-or-less steady population.
>
> The file gets reorganised once per w

. The processes run almost 15x7.
>
> Environment - SCO unix.
>
> Any memories stirred?
>
> Michael

Sponsored Links