dynamic memory error

[Cindy Lomas]

Is there a system parameter that can be changed to elimiate the following error?

out of dynamic memory in yy_scan_buffer

Thanks,
Cindy

[Uwe Zessin]

So-called dynamic memory is configured by the system parameter PAGEDYN, but the error message does not look like a standard OpenVMS message. Can you provide a bit more details about the application and an exact representation of the message?

[Cindy Lomas]

I am running a Cognos Powerhouse QTP program that is updating a table in an an Oracle database.

I agree the message does not look like a standard VMS error message. This program was converted from using an RMS indexed file to an equivalent Oracle table. So, basically the logic is the same. I've tried using different COMMIT points but no luck. The error message I've posted is exactly as it appears in the log and then the next line shows %NONAME-E-NOMSG, Message number 00000002 which really gives no other clues as far as I'm concerned.

Cindy

[Uwe Zessin]

Oh, OK. I really love this programming style :-(

Can you run AUTOGEN until the GENPARAMS phase and check the report if it wants to adjust PAGEDYN?

[Hein van den Heuvel · Accepted solution]

"dynamic memory" under VMS is mostly associated with 'the pool', as per prior replies.

However, I suspect that this program is really talking about a MALLOC failing, and that the underlying, but well hidden VMS error is really INSVIRMEM.
In which case you do not want to look at sysgen params (first) but at process environment. Check out:

$help/mess /fac=lib insvirmem

Most likely cause: PGFLQUO for the user.

Can you monitor (SHOW PROC/CONT) the process before it dies? How about Virtual memory and working set used?

Hein.

[Andy Bustamante]

Is this VAX or Alpha, and which version of OpenVMS?

If this is a process level quota, PGFLQUOTA is the most likely suspect, this is bound by VIRTUALPAGECNT in earlier versions of VMS.

[Cindy Lomas]

Thanks to everyone's suggestions. The problem was solved by increase the PGFLQUO for the user. I'm not sure how much it needed to be increased. It was at 150000 and I increased it to 300000. If it's unnecessarily high, hopefully it won't have a negative impact on anything.

Cindy

[Hein van den Heuvel]

Well, you could monitor the now working process to see how much is actually used.
But, this parameter is a safety barrier, like the maximum luggage allowance on for flight. Just making the number bigger, does nto make you consume more, it allows you to comsume more. And just like that airplane the limit can be different for individual passengers. And it wouldn't really matter if one passenger exceeded it some, but if all passenger exceed it, then you get into trouble. In the VMS case you would run out of pagefile. That's the underlying shared resource that needs protection, because if you run out, it is too hard to decide which piece of luggage is the least important. All you can do is refuse any more irrespective of the importance.

So... once you get this working in production, care to report back with the pro's and con's of replacing the indexed file with an oracle table? performance? resources? flexibility? re-orgs? backups? data useable from more applications? transaction semantics?...

Cheers,
Hein.

[Cindy Lomas]

Hein,

Thanks for your comments. So, I won't worry about the new value of 300000.

We are converting our whole system over to Oracle.

The main benefit I've seen so far is that we can give users access to Oracle through ODBC drivers allowing them to use the data in Excel etc.

Oracle will also allow us to move forward with web development and PC applications.

Cindy

[Ian Miller.]

note that there are ODBC drivers for RMS files available.