- From: olga <olga@goliath.eai.com>
- Date: Thu, 14 Jan 1999 16:17:08 -0600 (CST)
- To: www-lib@w3.org
Hi, I am having problems with -902 status as well. As I traced is down it always happens when one of the hosts closes connection after being idle for a while. Here is a trace: .... Looking up raven DNS Add..... `raven' with 1 home(s) to 100671c0 ParseInet... as port 80 on 10.11.1.10 with 1 homes HTHost 10069818 going to state TCP_NEED_SOCKET. Socket...... Created 10 Net Manager. Increasing active sockets to 2, 1 persistent sockets Socket...... Turned off Nagle's algorithm Socket...... Non-blocking socket Channel..... Hash value is 10 Channel..... Added 100675a8 to list 10058a48 Reader...... Created reader stream 1007edf8 Socket...... TCP send buffer size is 61440 for socket 10 Contacting raven HTHost 10069818 going to state TCP_NEED_CONNECT. HTDoConnect. WOULD BLOCK `raven' Event....... Register socket 10, request 1003acd0 handler 4259bc type HTEvent_CONNECT at priority 20 Event....... Registering socket for HTEvent_CONNECT Timer....... Dispatch timer 1003b5b8 Host Event.. host 10049f70 `skyblue' closed connection. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ !!!!!!! Event.. Entering HTEventList_unregister Event....... No more events registered for socket 9 Event....... Socket 9 unregistered for HTEvent_READ Event.. Entering HTEventList_unregister Event....... Couldn't find socket 9. Can't unregister type HTEvent_WRITE Channel..... Delete 10048e70 with semaphore 0 Socket read. FREEING.... Socket write FREEING.... Net Manager. Decreasing active sockets to 1, 1 persistent sockets Channel..... Deleted 10048e70, socket 9 Net Manager. 1 active sockets, decreasing persistent sockets to 0 Host info... removed host 10049f70 as persistent Event Loop.. calling select: maxfds is 10 Event Loop.. select returns -1 Event Loop.. select returned error 9 Event Loop..: Bad file number ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ Here select failes!!!! Event....... Dumping socket events soc event: pri millis callback param request timer: millis expires ? param callback 10 writ 1006f570: 20 -1 4259bc 10069818 1003acd0 WWWLibTerm.. Cleaning up LIBRARY OF COMMON CODE Net Object.. Kill ALL Net objects!!! Interrupted! Net Object.. Killing 10068688 Error....... Add 59 Severity: 1 Parameter: `Unspecified' Where: `HTLoadHTTP' HTTP Clean.. Called with status -902, net 10068688 Interrupted! Net Object.. Delete 10068688 and call AFTER filters Host info... Remove 10068688 from pipe Host Object. closing socket 10 Channel..... Semaphore set to 0 for channel 100675a8 Event.. Entering HTEventList_unregister Event....... Socket 10 unregistered for HTEvent_READ Event.. Entering HTEventList_unregister Event....... No more events registered for socket 10 Event....... Reset MaxSock from 10 to 1 Event....... Socket 10 unregistered for HTEvent_WRITE Channel..... Delete 100675a8 with semaphore 0 Buffer...... ABORTING... Socket read. FREEING.... Socket write FREEING.... Net Manager. Decreasing active sockets to 0, 0 persistent sockets Channel..... Deleted 100675a8, socket 10 Host info... removed host 10069818 as persistent Net Object.. Remove object 10068688 Response.... Created 10036aa8 Net After... calling 41746c (request 1003acd0, response 10036aa8, status -902, context 0) Request..... Delete 1003acd0 Request..... Deleting dangling output stream Net After. Delete All filters Net Before. Delete All filters Response.... Delete 10036aa8 ///////////////////////////////////////////////////////////////// In CHChannl : free_channel(..) ch->sockfd is set to INVSOC. /* Close the socket */ if (ch->sockfd != INVSOC) { NETCLOSE(ch->sockfd); /* HTEvent_unregister(ch->sockfd, all options); */ HTNet_decreaseSocket(); if (PROT_TRACE) HTTrace("Channel..... Deleted %p, socket %d\n", ch,ch->sockfd); ch->sockfd = INVSOC; Could this cause the problem? Maybe the socket should be deleted from the fd_set also before calling select on this set? Thanks, Olga Antropova.
Received on Thursday, 14 January 1999 17:13:27 UTC