IBM Mainframe Forum

Dear all,

I have a small snipet to capture the syslog using the sdsf rexx (ISFLOG) for 2 days ideally. The log content was around 8,00,000 lines. When i tried executing the script i met with error saying


Can anyone throw some light on this. I believe this was because of the fact that my tso logon id had no enough memory allocated (4K). Apologize if I or my thoughts are wrong.

Regards,
Viswa

Viswanathchandru wrote:Dear all,

I have a small snipet to capture the syslog using the sdsf rexx (ISFLOG) for 2 days ideally. The log content was around 8,00,000 lines. When i tried executing the script i met with error saying

Select all

 Error running SYSLOG1, line 17: Machine storage exhausted .

Can anyone throw some light on this. I believe this was because of the fact that my tso logon id had no enough memory allocated (4K). Apologize if I or my thoughts are wrong.

Regards,
Viswa

Presumably you are trying to suck the entire log into memory at once with "EXECIO * DISKR". Have you considered reading it a line at time using "EXECIO 1 DISKR" (or better yet, since EXECIO's performance is execrable, using a compiled language to do what you want)?

Hello Akatsukami,

Thanks for your time! Yes, reading some 1000 records at a time and writing it will work i guess. But I can't understand this.

or better yet, since EXECIO's performance is execrable, using a compiled language to do what you want)?

. Thanks!!

Regards,
Viswa

Issue PROFILE MSGID command, then run your rexx program. It should give you a message number that you can look up at LOOKAT.

Viswanathchandru wrote:Hello Akatsukami,

Thanks for your time! Yes, reading some 1000 records at a time and writing it will work i guess. But I can't understand this.

or better yet, since EXECIO's performance is execrable, using a compiled language to do what you want)?

. Thanks!!

You get wretched performance from EXECIO.

Write a Rexx script that reads a few crores of records from some data set. Write a COBOL or PL/I program that does the same. Use your shop's sort product to do likewise. Run all three in batch. You will find that the CPU time used (and probably elapsed time, although many external factors influence that) is much the highest in the Rexx script. The sort product wil almost certainly be lowest. The HLL program will be intermediate, although probably closer to the sort.

execrable

I learned a new word.