I tested some UI enhancement (loading state 'hack' in current gwt celltable 
version) 
in case of cold start so that app user knows what is going on. To test 'cold 
start' I waited for every instance to stop (it would be nice feature to be 
able to stop them manually from dashboard) and then did some UI action in 
gwt application.

First I noticed that latency is much higher than expected for our request 
cold start. Instead of around 5000ms it was several times higher. Then I 
looked into a log and noticed pending_ms time for cold start and it happened 
before.

If you look at my gae 
log<http://picasaweb.google.com/matija.jerkovic/GAEStrangeColdStart?authkey=Gv1sRgCNaa1IPpsuz-8AE#5554200338544678018>
 you 
can see that latency was 20480ms with 4815 cpu ms and pending_ms=9795, but 
this request appstat 
data<http://picasaweb.google.com/matija.jerkovic/GAEStrangeColdStart?authkey=Gv1sRgCNaa1IPpsuz-8AE#5554200339962305794>
 has 
event more strange data. What is now real 6255 ms ?

I assume that math behind this numbers is:
20480 = 9750 + 6255 + 4815 - 340 ms (as part of regular non-cold start 
request cpu usage, because 4815 includes it with cold start initialization)

Can somebody explain me this numbers and why my cold start request has 
pending_ms time if this is first instance creation and there was no other 
activity ? 

-- 
You received this message because you are subscribed to the Google Groups 
"Google App Engine" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/google-appengine?hl=en.

Reply via email to