[jira] [Commented] (TS-1935) ERR_UNKNOWN/000 showing up 10.4% of the time in production logs

["Bryan Call (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/TS-1935?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13802332#comment-13802332 ] 

Bryan Call commented on TS-1935:
--------------------------------

Still seeing it on 4.0.1:

-bash-4.1$ logcat squid.blog | cut -f 4 -d ' ' | sort | uniq -c | sort -rn | head
2960322 TCP_HIT/200
 549929 TCP_IMS_HIT/304
 389811 ERR_UNKNOWN/000
  88349 TCP_MISS/200
  32607 TCP_IMS_HIT/200
   9694 ERR_CONNECT_FAIL/404
   6711 TCP_REFRESH_HIT/200
   6672 TCP_HIT/302
   6343 TCP_HIT/404
   6106 TCP_HIT/400

> ERR_UNKNOWN/000 showing up 10.4% of the time in production logs
> ---------------------------------------------------------------
>
>                 Key: TS-1935
>                 URL: https://issues.apache.org/jira/browse/TS-1935
>             Project: Traffic Server
>          Issue Type: Bug
>            Reporter: Bryan Call
>             Fix For: 5.0.0
>
>
> Need to investigate in what conditions this happens.
> * Keep-alive or not
> * If any requests have happened on the connection
> * Is the client or the server closing the connection at the time
> * Verify no bytes have been sent from the client, looking at the code it looks like it
> Ideally if the client has not sent any bytes (has not attempted to make a request) then we shouldn't log the error message.  If it is possible a state machine should not be created in this scenario.
> related bug:
> TS-1056
> related checkin:
> 425e3d7dcb16b40075c7678e71d00b2fcb4b4705
> I also see this happening in on my forward proxy using Chrome:
> [bcall@homer trafficserver]$ traffic_logcat -f squid.blog | grep 000
> 1370281253.174 0 192.168.1.30 ERR_UNKNOWN/000 0 - / - EMPTY/- - -
> 1370281253.450 0 192.168.1.30 ERR_UNKNOWN/000 0 - / - EMPTY/- - -
> 1370281253.891 0 192.168.1.30 ERR_UNKNOWN/000 0 - / - EMPTY/- - -



--
This message was sent by Atlassian JIRA
(v6.1#6144)