Conference iosg::all-in-1_v30

324.0. "ALL-IN-1 CPU RESPONSE TIME" by AIMTEC::MIXON_C () Wed Mar 25 1992 16:21

    I had a user write to the ALL-IN-1 support center with the following
    question:
    
    If anyone happens to know offhand if these CPU times are high and if
    when you do an exit does it continue reading the daf entry, I would
    appreciate a response.  
    
    Thank you!
    Cindy

Hi,

we had an interesting occurance today and I was wondering if you could explain
what happened (in gory detail please...)

a user sent a mail message to 1291 users (not via :SUBSCRIBERS), all 6 nodes in
the cluster (2x8800, 2x6430, 1x6510 and 1x6520) all became CPU bound.  In doing
a little research on the problem, it appears that even though users were
hitting exit screen to get out of reading the entire distribution list, they
were still consuming ~30 seconds of CPU time to "exit" the document.

Some CPU figures...

Before read                              - 1 Minute 13.57 seconds of CPU time
After first screen of message appeared   - 1 Minute 18.50 seconds of CPU
after getting to start of DL and {key 0} - 1 Minute 18.57 seconds of CPU
Returned to EM menu                      - 1 Minute 47.90 seconds of CPU

So, setup time is ~5 CPU seconds
    read message time is ~.1 seconds
    exit time is ~30 CPU seconds

I've been using ALL-IN-1 since V1.2 days and these times seem extraordinarily
long to me...  It almost appears that even though I've told it to get out it
still insists on reading ALL of the records in SDAF associated with this
message.

This is on VMS 5.4-1A, ALL-IN-1 V2.4 with K603 installed, 8800 processor

    Since I have RMS write access to the conference, I have reposted .0
    removing the customer's phone number in a more disk space efficient
    way!
    
    Graham
    
    PS My guess would be that we had the SDAF open and reading ahead down
    the address lists or something?????