From: A. <rem...@ya...> - 2006-01-11 12:59:00
|
Hi all I have no more SIGSEV after solving my problems with invalid utf8 events, but I found that I have now SIGABRT coming after some times. I was able to pinpoint the problem to the fact that osync_member_commit_change received NULL as value of its change parameter. It's clearly visible in TRACE 1. >From a debug session, it seems that the message is coming from the file_sync plugin with a NULL payload. Since the problem was related to a message, I tried to follow that message in the log files. I found out that message 0xb310ebe0 had a problem (see trace 2) The message is printed when the function timeoutfunc is called (file osengine_queue.c). In the same function, there is a call to itm_message_move_data, that is reseting the payload of the message in timeout while moving it data to the error message. After that call, timeoutfunc is setting the message status to ANSWERED thus, I believe, allowing the handler to deal with that message, and then aborting because of the NULL payload. Since I don't want to broke something in the engine and the queue management, I need advice before trying to propose a modification. Armin ? Regards RemyA PS: I just saw the news about the conf. Good trip guys, and make us a shining new release :-) TRACE 1: [1136979753.739199] >>>>>>> client_message_handler(0x8839810, 0xb310ebe0, 0x8833808) [1136979753.739270] [CLI] DEBUG: Client message handler called for message "COMMIT_CHANGE" [1136979753.739428] >>>>>>> osync_member_commit_change(0x8832320, (nil), 0xc16c14, 0xb310ebe0) TRACE 2 [1136979753.733361] [ENG] ERROR: Timeout while waiting for a reply to message 0xb310ebe0:"COMMIT_CHANGE". Sending error 0xb32f8c70 -- E-mail : Rem...@ya... Yahoo Architecture Europe (http://www.yahoo.com/) Yahoo Id: kelkooremya |