I use cvs 20090114 (pre meta data) with sqlbox 0.72, tested against 100 thousands MO, speed around 100 MO/sec. Still fine.
Willy Sent from my Sony Ericsson XPERIA X1. -----Original Message----- From: Alejandro Guerrieri <alejandro.guerri...@gmail.com> Sent: 29 December 2009 18:41 To: Willy Mularto <sangpr...@gmail.com> Cc: ken.bell...@gmail.com; users@kannel.org; de...@kannel.org Subject: Re: SQLBox Issues and Limit per cycle SQLBox patch You asked for it, you've got it ;) [Sorry, I'm cross-posting this one, since it's probably people interested on both lists for different reasons] http://www.kannel.org/~aguerrieri/SqlBox/Snapshots/sqlbox-meta-20091229.tar.gz This snapshot is not commited to CVS yet, since it needs proper testing on ALL DB engines: mysql, postgres, sqlite(2), sqlite3, oracle, mssql/sybase and sdb. Volunteers are needed to test all of them (I'll do my best as well, but any help is appreciated). To build it, run ./bootstrap first, then ./configure --with-<dbengine>, make and sudo make install. You need a Kannel compiled with support for the same DB engine you're going to use of course. WARNING: There still seems to be an issue with MO under heavy load: some messages are lost. After further testing, this doesn't seem to be related with meta-data at all. To reproduce the error, you need to put sqlbox between bearerbox and smsbox and use it to capture all traffic (MO and MT). Then, if you sent a bunch of messages from a fake smsc at high rates (100 or more messages with .01 delay is enough to cause it), some of them are not recorded nor forwarded to smsbox. I need to investigate it further, I've only tried with mysql so far, so I'm not sure if this is a DB problem, or something related with threading or so. Actually I think the same problem would happen with a pre-meta-data kannel version (anyone can confirm this?). NOTE: There's NO problem with MT, if you use SqlBox to send messages by inserting on send_sms, then you won't notice any problems at all. Last but not least, I'm out of luck building on Snow Leopard. It compiles but segfaults afterwards. [The entire original message is not included]
