 |
|
|
 |
| Subject: Extended HTTP log response time |
 |
 |
 |
| Product Area: Administration |
 |
| Technical Area: Administration |
 |
| Platform: Windows |
 |
| Release: 8.5.1 |
 |
| Reproducible: Not applicable |
 |
 |
 |
 |
We have extended HTTP logs writing to a text file, I have question about one of the outputs, rsponse time.
The IBM definitio is: Amount of time, in milliseconds, to process the request, so my rewquest below took 78ms (I reordered the cols from the "normal" colum output).
xx.x.x.xx CN=XXXX XXXXXX/O=XXXX [09/Feb/2012:07:30:10 200 1320 78 GET /XXXXXX.nsf/someurlreqwuest HTTP/1.1
Does this mean: server received the request at 07:30:10 and it was processed and a response was sent out 78ms later or, server received request at some time prior to that but was so busy...maybe totol HTTP thread saturation...and 30 secs later was able to process the request, so from user standpoint, request took 30secs + 78ms.
We've done some things with the HTTP logs and pivot tables, so we can see HTTP thread state accross periods of time, and it doesn't look like http threads are saturating. Is there a way to monitor HTTP thread state over a period of time, and if it hits the MAX, generate an event for DDM? I couldn't find one, and short of "tell http show thread state" at console (or maybe getting a program doc to do it periodically...but it might miss a MAX event), looking at stats (wmhich I can only get avs or MAXs from)...is there any way to monitor for http thread state maxing out and log the time?
Thanks in advance for any comments,
Nick
 
Feedback number WEBB8RBMNZ created by ~Holly Quetfookony on 02/09/2012
Status: Open
Comments:
|
| |
|
Return to top
Printer-friendly
