Re: Can't DISCONNECT|CLOSE when -439

In message <3F4507B7.9090302@reach.co.za>, Corne' Cornelius <cornec@reach.co.za> writes >Hi, > >Any ideas why a threaded application can't > >DISCONNECT :connection or CLOSE DATABASE > >when SQLCODE was -439 before a sqlbreak(), and 0 after the sqlbreak() ? > >it just encounters another -439 error, even if sqlbreak() is called before. > >sqlbreakcallback(0L,callback_func) doesn't even call the callback_func >as it's supposed to. > My manual says that if the timeout is 0, then the callback function executes immediately, and then 'over and over again'. However it also says that the callback function will never be called if you use a shared memory connection. I guess the application is correctly built threaded.. >maybe a socket timeout ? would a signal(SIGPIPE, SIG_IGN) help ? >onstat -g ses still shows the connection though> >maybe a bug in the engine ? how could i get around this ? > >Thanks >Corne' > >!Exclude Disclaimer! > -- Andrew Lennard andy@kontron.demon.co.uk

The callback functions wasn't even called when i used 0 as timeout. Maybe the client loses the signal handler, or gets blocked ? Not using shared memory connection. using onsoctcp The only thing that seems to work, is to kill any thread that got the -439 possible runaway process, and create a new thread. Andy Lennard wrote: > In message <3F4507B7.9090302@reach.co.za>, Corne' Cornelius > <cornec@reach.co.za> writes > >>Hi, >> >>Any ideas why a threaded application can't >> >>DISCONNECT :connection or CLOSE DATABASE >> >>when SQLCODE was -439 before a sqlbreak(), and 0 after the sqlbreak() ? >> >>it just encounters another -439 error, even if sqlbreak() is called before. >> >>sqlbreakcallback(0L,callback_func) doesn't even call the callback_func >>as it's supposed to. >> > > > My manual says that if the timeout is 0, then the callback function > executes immediately, and then 'over and over again'. > > However it also says that the callback function will never be called if > you use a shared memory connection. > > I guess the application is correctly built threaded.. > > >>maybe a socket timeout ? would a signal(SIGPIPE, SIG_IGN) help ? >>onstat -g ses still shows the connection though>> >>maybe a bug in the engine ? how could i get around this ? >> >>Thanks >>Corne' >> >>!Exclude Disclaimer! >> > >