[Varnish] #320: ESI fails with POST fails

[Varnish]

#320: ESI fails with POST fails
----------------------+-----------------------------------------------------
 Reporter:  kierank   |        Owner:  phk     
     Type:  defect    |       Status:  reopened
 Priority:  normal    |    Milestone:          
Component:  varnishd  |      Version:  trunk   
 Severity:  normal    |   Resolution:          
 Keywords:            |  
----------------------+-----------------------------------------------------
Comment (by phk):

 I've looked over the logs, and I can reproduce the scenario but not the
 failure.

 It looks to me like the communcation with the backend on fd=11 fails
 during the GET attempt for the included object and therefore we go from
 TxHeader to BackendClose in line 148-149.

 The most likely explanation is that the backend closed the connection and
 we didn't know it.

 To see if this is the case, you could make the following experiment to
 prevent reuse of the
 TCP connection:

 Insert an explicit "Connection: close" header in the rereq in the
 vcl_pass{} function.

 If that helps, we may simply be incredibly unluck about a timing race with
 your backend.

 Poul-Henning

-- 
Ticket URL: <http://varnish.projects.linpro.no/ticket/320#comment:7>
Varnish <http://varnish.projects.linpro.no/>
The Varnish HTTP Accelerator