Some news.

I created a small test-case that submit a request every time bigger and I've been able to reproduce the problem. The strange thing is that the more logs I add the more loops are performed without errors. It makes me think that is something related to the data sent from apache to the input stream and that in some way are lost or not read correctly. Looking at the request processing I can see that the previous working requests works like that (this is mod_mono log):

*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 32768, actual size == 32768
received from apache == 2520
received from apache LE == 2520
write size 2520 - 4
status 4
received from apache 2 == 2520
write string 2520
status 2520
*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 32768, actual size == 32768
received from apache == 2520
received from apache LE == 2520
write size 2520 - 4
status 4
received from apache 2 == 2520
write string 2520
status 2520
*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 32768, actual size == 32768
received from apache == 3780
received from apache LE == 3780
write size 3780 - 4
status 4
received from apache 2 == 3780
write string 3780
status 3780
*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 31078, actual size == 31078
received from apache == 2528
received from apache LE == 2528
write size 2528 - 4
status 4
received from apache 2 == 2528
write string 2528
status 2528
*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 28550, actual size == 28550
received from apache == 2520
received from apache LE == 2520
write size 2520 - 4
status 4
received from apache 2 == 2520
write string 2520
status 2520

.... and so on until ....

*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 3350, actual size == 3350
received from apache == 3350
received from apache LE == 3350
write size 3350 - 4
status 4
received from apache 2 == 3350
write string 3350
status 3350
*****************************************
Command received: SET_STATUS (8)
Command received: SET_RESPONSE_HEADERS (2)
Headers length: 109
Command received: SEND_FROM_MEMORY (0)
sending from memory without flush
Command received: CLOSE (4)
Done. Status: 0

Usually xsp reads from mod_mono the same value that mod_mono reads from apache and everything works. When the request fails I see:

Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 32768, actual size == 32768
received from apache == 32768
received from apache LE == 32768
write size 32768 - 4
status 4
received from apache 2 == 32768
write string 32768
status 32768
*****************************************
Command received: GET_CLIENT_BLOCK (7)
*****************************************
requested size == 32768, actual size == 32768
received from apache == 8277
received from apache LE == 8277
write size 8277 - 4
status 4
received from apache 2 == 8277
write string 8277
status 8277
*****************************************
Command received: SET_STATUS (8)
Command received: SET_RESPONSE_HEADERS (2)
Headers length: 107
Command received: SEND_FROM_MEMORY (0)
sending from memory without flush
Command received: CLOSE (4)
Done. Status: 0

It seems ok but when mod_mono read 8277 bytes from apache xsp reads a value of 1465542519 and it throws the Houston error. There's someone that understand these kind of things that can drive me to the correct road to find the solution?
Thanks in advance.



_______________________________________________
Mono-devel-list mailing list
Mono-devel-list@lists.ximian.com
http://lists.ximian.com/mailman/listinfo/mono-devel-list

Reply via email to