Summary:	ASTERISK-00929: [patch] chan_sip fails into setting the ipaddr of the mysql peer (sometimes)
Reporter:	Matteo Brancaleoni (mbrancaleoni)	Labels:
Date Opened:	2004-01-28 05:52:51.000-0600	Date Closed:	2004-09-25 02:55:32
Priority:	Minor	Regression?	No
Status:	Closed/Complete	Components:	Core/General
Versions:		Frequency of Occurrence
Related Issues:
Environment:		Attachments:	( 0) chan_sip.patch
Description:	sometimes (like 1 call out of 10) a very strange behaviour happens using mysql friends: in sip_create_addr there's a call to mysql_peer that retrieves the data from the db and set peer addr, among others. but returning to sip_create_addr the ip address is 0.0.0.0 ! I mean... doing debug ater inet_aton(rowval[x], &p->addr.sin_addr); (line 730 into chan_sip) the ip addr is *always* correct... then the functions return to line 794... and here sometimes the p->addr.sin_addr is 0.0.0.0 ,even if before was correct... there must be something wrong, but I wasn't able to find out where. I made that quick patch that uses a temp var to store the ipaddr and all works ok (just to make the addr assignment into create_addr and not mysql_peer). the same routines into chan_iax2 doesn't have this problem at all. any idea?
Comments:	By: Olle Johansson (oej) 2004-01-28 15:28:53.000-0600 WHEN does this happen? At registration? Or dial out to peer? mysql_peer sets dynamic=1 Hmmm. This is ugly. Please explain more. By: Matteo Brancaleoni (mbrancaleoni) 2004-01-28 15:42:18.000-0600 * registrations : always ok, I can register without problems and on the db the values are stored correctly. that happens when I dialout to the peer, but not always (like above 1 call out of 10) and is more regular if I wait at least 20 secs or more after the last succesfull call. the problem is on the peer side... Here's the scenario: * dialout to peer (I assume that call is ok) -> case A: hangup and quickly redial... no probs. I can always dial the peer -> case B: hangup, wait some time (30secs ?)... dial again... error.   on __sip_xmit I see WARNINGs with IP addr == 0.0.0.0 so I did some debug. since with mysql friends, every peer is dynamic and temponly, every time a sip peer is dialed sip_create_addr calls mysq_peer to get peer data (the peer is registered, so I have all data, like ipaddr & port on the db). So I added 2 ast_logs: first in mysql_peer, dumping the ipaddr, after the assignment to the struct second in sip_create_addr, just after mysql_peer is called, dumping the same value (ip addr). They must be the same. when all is ok , they're the same. when I get error, the first is ok, the second one is 0.0.0.0 ! but, if I use a temp var like a *char[20] into the peer struct, where I save the string from the db in mysql_peer and in sip_create_addr I do the real assignment.... all is ok. (see my patch). my workaround is solid, but still I cannot get why the ipaddr is lost somewhere going back from mysql_peer to sip_create_addr... By: Rob Gagnon (rgagnon) 2004-02-23 16:01:14.000-0600 isn't the 0.0.0.0 simply being caused by an expired registration check following the read of data from the table? Current cvs 1.299 shows: if ((nowtime - regseconds) > default_expiry) memset(&p->addr, 0, sizeof(p->addr)); following the read of the data... This problem might be fixed by bug ASTERISK-1011052 edited on: 02-24-04 11:10 By: z_smurf (z_smurf) 2004-02-23 16:48:07.000-0600 Yes, this is a fixed in ASTERISK-1011052. By: Malcolm Davenport (mdavenport) 2004-02-24 15:02:13.000-0600 Changes made in CVS.  Thank you. :)