Re: Loading even small files reqs. clicking in/out multi. times
>>>> "David" == David M Karr
<dmkarr(a)earthlink.net> writes:
>>>> "Andy" == Andy Piper <andyp(a)bea.com>
writes:
Andy> At 01:58 PM 6/21/2002 -0700, David M. Karr wrote:
>> Well, I'm not certain. If the output of
"list-processes" really shows all of
>> your sub-processes, then I was probably still under 64. I probably had 52-53
>> processes. However, I'm a little uncertain exactly what entries in that
list
>> signify. I had thought that if I execute '(kill-process
"WWW<72>")' it would
>> directly kill that process. However, this gives me an error indicating "no
>> such process" (I don't remember the exact error message). So, either
the
>> "Proc" column doesn't represent a process name you can send to
"kill-process",
>> or many of those entries don't represent real processes.
Andy> I guess it would be helpful to know if its a particular type of process that
Andy> causes the problem or the number of processes or a combination of both.
David> Ok, next time it hits, I'll recheck the symptom after deleting each WWW
buffer.
David> I'll also probably try deleting shell buffers one by one, checking the
symptom
David> each time. I think I might try manually creating a large number of shell
David> buffers, to see if that causes it.
Ok, I was in a state where the symptom was not occurring. I recorded a
keyboard macro that created a new shell buffer, and I executed it 50 times.
This took quite a while to finish. After a while I discovered that I could get
it to finish faster by clicking out of the frame, and back in, just like with
loading files. After it finished creating the shell buffers, I had the symptom
back. I also noticed that the CPU was pegged all throughout this, after I
started to create the shell buffers.
I then went through the process of deleting a shell buffer and checking the
symptom. I did this until I got down to about 30 processes. At that point the
symptom went away (and CPU usage went down to normal). However, I then
couldn't get the symptom back by adding back more shell buffers.
I wonder if this problem has something to do with creating numerous processes
in a very short time period. Whenever I've seen this happen with the "WWW"
processes, they're all to the same URL, so they were likely from the same email
message, so those connections were probably all created at just about the same
time.
--
===================================================================
David M. Karr ; Java/J2EE/XML/Unix/C++
dmkarr(a)earthlink.net