[Asterisk-bsd] * Not working after upgrade

[Thomas Sandford]

"Frank Griffith" <glassdude45 at yahoo.com> wrote:
> Kim Culhan <w8hdkim at gmail.com> wrote:
KC> If this doesn't work it would probably be best to do a fresh
KC> installation of FBSD and take it from there.
>
FG> This is what sucks about Windows. And it's hard to believe I had to
FG> resort to such tactics to clear the cobwebs with FreeBSD.
>
KC> In the case of Windows its not unusual for settings to be made in the 
registry
KC> by installed programs with no 'easy way' to trace their location or the 
registry
KC> values which were set.
>
KC> This is not the case with FreeBSD.
>
KC> The fact there were files in /usr/local/lib/asterisk/modules left over 
from
KC> the first installation attempt is a design 'feature' of the Asterisk
KC> build system
KC> and not something related to the FreeBSD operating system or the 
Asterisk port.
>
KC> When the Asterisk port is installed, there are not many places where
KC> files are written so 'clearing cobwebs' can be fairly easilly done
KC> when the 'webs
KC> are known to be coming from the Asterisk port itself.
>
KC> If you read the archives of this list, I think you will agree that in
KC> most cases when
KC> things don't seem to be working the way they should, the cause has been
KC> traced to factors other than a defective port or a flaw in FreeBSD.
>
FG> That's all true...but what else is also true was all I did was this:
>
FG>  portsnap fetch ; portsnap extract
>
FG>  cd /usr/ports/net/asterisk
FG>  make deinstall
FG>  make install clean
>
FG>  Things went pretty much downhill from there. But thanks to everyone who 
offered advice. And I now have the latest version of FreeBSD-RELEASE on my 
system.

First of all, as the person who fundamentally "did" the upgrade of the port 
from 1.2.9.1 to 1.2.12.1 (though the actual commit was from the port 
maintainer, sobomax at freebsd.org, and he made some cosmetic changes to my 
original patch), I am still bemused as to why it didn't "just work" for 
Frank Griffith, and even more so as to why a reinstall of FreeBSD should 
have helped.

Having said that there certainly would be instances where:
% portsnap fetch
% portsnap extract
% cd /usr/ports/net/asterisk
% make deinstall
% make install clean
would not be sufficient. If everyone has got all the dependencies exactly 
right it _should_ be, but with Asterisk in particular there are subtle 
changes in dependencies between versions that are not always picked up - 
particularly if the person doing the port upgrades things themselves in the 
"wrong" order and so doesn't spot the dependency. (I got caught out on this 
myself recently when I updated libpri locally then set out to do an update 
on the asterisk port - not realising that the new asterisk needed the libpri 
update).

I can only assume there was a subtle variant of this problem that occured in 
this case, and that by installing over a brand new FreeBSD (and fresh ports 
tree) brand new versions of all dependencies were installed, fixing the 
original problem.

I'd actually partially disagree with
"KC> The fact there were files in /usr/local/lib/asterisk/modules left over 
from
KC> the first installation attempt is a design 'feature' of the Asterisk
KC> build system and not something related to the FreeBSD operating
KC> system or the Asterisk port.
"

If everything is done from ports there _shouldn't_ be anything like modules 
left over in /usr/local - that's part of what ports should be _doing_ There 
might be things like configuration files, logs, etc left over (indeed there 
should be, or upgrades would be a painful process), but if a port is 
installing executables, libraries, .so's, include files, etc onto the system 
and not removing them (or indeed putting them into /usr/local (or $PREFIX) 
at any time other than in the make install target) then the port is just 
plain wrong (and indeed will fail on the ports cluster "pointyhat" builds - 
and sooner or later get marked "BROKEN").

Anyway - I'm glad the OP's main problem is now resolved - possibly time for 
a new thread to discuss the warnings he (and others) is (/are) reporting.
-- 
Thomas Sandford