On 05/10/2011 02:08 PM, John Craws wrote:
> 5. At some point (> preview size, < total size) Squid sends a "TCP
> Window Full" to the origin server.
> 6. Squid and the origin server enter a TCP Zero Window Probe cycle
> that they never get out of.
I am glad you found a solution, but there could be a Squid bug lurking
here anyway.
I did not quite realize Squid was doing Preview here. After the ICAP 200
response to Preview, Squid should have enabled auto-consumption of the
origin server response, which would have prevented the Squid-origin
server transfer from getting stuck.
I did not check the code, but based on your description alone, it sounds
like Squid did not enable BodyPipe auto-consumption. It would be nice to
check and, if needed, fix that.
Cheers,
Alex.
Received on Tue May 10 2011 - 21:26:47 MDT
This archive was generated by hypermail 2.2.0 : Wed May 11 2011 - 12:00:04 MDT