On 03/07/2006 10:10 PM, Jim Jagielski wrote: > William A. Rowe, Jr. wrote: > >>For the benefit of folks here and not on mod_jk... >> > > > Check out jk_ajp_common.c in tomcat-5.5-connectors/jk/native/common > where whether the flush is forced for each JK_AJP13_SEND_BODY_CHUNK > is controlled by the FlushPackets. Also note how we vary > behavior at JK_AJP13_END_RESPONSE :) > > This is what we need to add to mod_proxy_ajp... well, I > mean the "forcing" is already there, but we need to make > it runtime. I need to look at how m_p_a handles the > JK_AJP13_END_RESPONSE state, since I think we have a problem > there as well.
We sent an eos bucket up the chain in this case, so the core output filter will flush. I am not sure if it tries a flush if no data is there to flush (the thing that is prevented at JK_AJP13_END_RESPONSE) Regards Rüdiger
