[jira] [Comment Edited] (TS-2998) sscl is occasionally zero

["Sudheer Vinukonda (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/TS-2998?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14119102#comment-14119102 ] 

Sudheer Vinukonda edited comment on TS-2998 at 9/3/14 1:05 AM:
---------------------------------------------------------------

Now that I think of it, this may be related to TS-3049. Scott, do you know if you are seeing this issue only for a spdy client connection or do you also see this for non-spdy clients? Now that we have the fix for TS-3049 available, it might also be interesting to see if the issue still occurs. 


was (Author: sudheerv):
Now that I think of it, this may be related to TS-3049. Scott, do you know if you are seeing this issue only for a spdy client connection or do you also see this for non-spdy clients?

> sscl is occasionally zero
> -------------------------
>
>                 Key: TS-2998
>                 URL: https://issues.apache.org/jira/browse/TS-2998
>             Project: Traffic Server
>          Issue Type: Bug
>          Components: Logging
>            Reporter: Scott Beardsley
>             Fix For: sometime
>
>
> Running ATS 5.0 it looks like sscl in the extended2.log is occasionally zero even for cache misses and HTTP 200 reponses...  Here is how I am finding these cases:
> tail -f extended2.log|awk '$23~/TCP_MISS/&&$11~/200/&&$12~/^0/{print $10,$11,$12,$23,$7}'
> 21452 200 0 TCP_MISS http://search.example.com/search?p=pizza
> You can see that pscl ($10) is non-zero but sscl ($12) is zero. Where is ats getting the content from if it isn't coming from the origin? Is this related to the value of the content-length header from the origin? Ideally we would want to see the raw number of bytes coming from the origin.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)