VoIP Mailing List Archives

[nik.middleton at noble...]

Well I can only assume build 13537 is brain dead. Surely I shouldn’t have to edit a whole bunch of configs to get it working. FS now takes 3 minutes to start, with no indication as to what it’s looking for in the logs. That said, to date ‘make current’ has always worked well for me. Guess I was bound to hit a bad one eventually.

Still, it’s very frustrating.


From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Brian West
Sent: 01 June 2009 22:29
To: freeswitch-users@lists.freeswitch.org
Subject: Re: [Freeswitch-users] Make current fails


I can tell you how to fix it but it'll cost ya!


/b





Spoke too soon.



Clean compile and install, but now FS hangs for about 5 mins on startup



Error [unterminated ${var}] in file /usr/local/freeswitch/conf/autoload_configs/../jingle_profiles/client.xml line 12

Error including /usr/local/freeswitch/conf/autoload_configs/../mrcp_profiles/*.xml (Invalid or incomplete multibyte or wide character)



The first error is a typo in the sample, but the second error, I don’t have that DIR at all. I presume that this dir has been added, but how to I create these without overwriting my working configs?





Regards





Brian West

brian@freeswitch.org (brian@freeswitch.org)




-- Meet us at ClueCon! http://www.cluecon.com

[brian at freeswitch.org]

NO its not a bad one at all. Its switch_nat_init(); in switch_core.c since your network must be eating the packets its sending out to detect if you're behind nat or not... and not getting an ICMP unreachable like it should be getting... the joys of admins that block all ICMP like idiots. ICMP has many uses... and outright blocking it is stupid. (This is my assumption cuz its what makes sense in this case)

So you're getting hit by the nice retry/timeout loop in the natpmp software we just added and possibly the upnp lib too.


So for now edit switch_core.c and comment out switch_nat_init();


I'm working my ass off to ensure that our users that do have to live in these insane nat scenarios can do so without much if any pain. Most of which uses SMB/Consumer grade routers which these two libs we added would allow us to poke holes and setup stuff and make it painless as possible.


Soon you'll have an option in switch.conf.xml to turn it off.


Please next time don't be so demanding and calling builds brain dead .. when in fact its trying to become more aware of its network config without much user input.


/b

On Jun 1, 2009, at 5:24 PM, Nik Middleton wrote:

[freeswitch at davidnic...]

Firstly, thanks for continuing to provide a superior piece of VOIP software. I have a couple of small and unrequested suggestions: 1: have "make current" after the svn update do ` (./configure && make) || (./boostrap.sh && configure && make)` instead of what it does now which is presumably the first piece only 2: have the uninstall target back up your current binaries before removing them, and include some info about restoring them, for us idiots who neglect to do such things as tar czf "freeswitch-bin-`date`.tgz" /usr/local/freeswitch/bin before charging ahead with our dependable "make current && make install" procedures. On Mon, Jun 1, 2009 at 10:33 PM, Brian West (brian@freeswitch.org) wrote:

[dave at 3c.co.uk]

At the risk of evisceration (but with the intention of helping avoid future brain dead build vs. idiot admin debates), I'd suggest that, when significant new bits are added to the switch core, they should default to being off and require a configuration option to turn them on. Such config options can be added to the default config; that way new installs will have the new functionality enabled by default, but those upgrading from an older install will need to enable them manually, reducing the risk of stuff breaking. --Dave ----- Original Message ----- From: Brian West (brian@freeswitch.org) To: freeswitch-users@lists.freeswitch.org (freeswitch-users@lists.freeswitch.org) Sent: Monday, June 01, 2009 11:33 PM Subject: Re: [Freeswitch-users] Make current fails (build 13537) NO its not a bad one at all. Its switch_nat_init(); in switch_core.c since your network must be eating the packets its sending out to detect if you're behind nat or not... and not getting an ICMP unreachable like it should be getting... the joys of admins that block all ICMP like idiots. ICMP has many uses... and outright blocking it is stupid. (This is my assumption cuz its what makes sense in this case) So you're getting hit by the nice retry/timeout loop in the natpmp software we just added and possibly the upnp lib too. So for now edit switch_core.c and comment out switch_nat_init(); I'm working my ass off to ensure that our users that do have to live in these insane nat scenarios can do so without much if any pain. Most of which uses SMB/Consumer grade routers which these two libs we added would allow us to poke holes and setup stuff and make it painless as possible. Soon you'll have an option in switch.conf.xml to turn it off. Please next time don't be so demanding and calling builds brain dead .. when in fact its trying to become more aware of its network config without much user input. /b On Jun 1, 2009, at 5:24 PM, Nik Middleton wrote: Well I can only assume build 13537 is brain dead. Surely I shouldn’t have to edit a whole bunch of configs to get it working. FS now takes 3 minutes to start, with no indication as to what it’s looking for in the logs. That said, to date ‘make current’ has always worked well for me. Guess I was bound to hit a bad one eventually. Still, it’s very frustrating. Brian West brian@freeswitch.org (brian@freeswitch.org) -- Meet us at ClueCon! http://www.cluecon.com _______________________________________________Freeswitch-users mailing listFreeswitch-users@lists.freeswitch.orghttp://lists.freeswitch.org/mailman/listinfo/freeswitch-usersUNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-usershttp://www.freeswitch.org

[brian at freeswitch.org]

Its coming soon!

/b

On Jun 2, 2009, at 6:23 AM, David Knell wrote:

[larclap at yahoo.com]

Brian,

I’m probably not the only one here, but much of what I have to do to get Freeswitch going is new to me. Never installed or really worked with Linux and scripting; just a little xml. It is challenging. Freeswitch is interesting, appealing and challenging. The work your group has done is amazing. Given this, interacting with you can be intimidating.

I am experiencing the slow start with build 13532. I assume that “block all ICMP” refers to the firewall/gateway. If this is correct, why is it that I can ping the firewall from the Freeswitch box? Can you explain in more detail what it might be on my network that is blocking ICMP? All my clients and Freeswitch itself are behind a NAT firewall.

Thanks Lars

Linux fs 2.6.18-128.1.10.el5 #1 SMP Thu May 7 10:39:21 EDT 2009 i686 i686 i386 GNU/Linux

From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Brian West
Sent: Monday, June 01, 2009 3:33 PM
To: freeswitch-users@lists.freeswitch.org
Subject: Re: [Freeswitch-users] Make current fails (build 13537)



NO its not a bad one at all. Its switch_nat_init(); in switch_core.c since your network must be eating the packets its sending out to detect if you're behind nat or not... and not getting an ICMP unreachable like it should be getting... the joys of admins that block all ICMP like idiots. ICMP has many uses... and outright blocking it is stupid. (This is my assumption cuz its what makes sense in this case)


So you're getting hit by the nice retry/timeout loop in the natpmp software we just added and possibly the upnp lib too.



So for now edit switch_core.c and comment out switch_nat_init();



I'm working my ass off to ensure that our users that do have to live in these insane nat scenarios can do so without much if any pain. Most of which uses SMB/Consumer grade routers which these two libs we added would allow us to poke holes and setup stuff and make it painless as possible.



Soon you'll have an option in switch.conf.xml to turn it off.



Please next time don't be so demanding and calling builds brain dead .. when in fact its trying to become more aware of its network config without much user input.



/b


On Jun 1, 2009, at 5:24 PM, Nik Middleton wrote:




Well I can only assume build 13537 is brain dead. Surely I shouldn’t have to edit a whole bunch of configs to get it working. FS now takes 3 minutes to start, with no indication as to what it’s looking for in the logs. That said, to date ‘make current’ has always worked well for me. Guess I was bound to hit a bad one eventually.



Still, it’s very frustrating.





Brian West

brian@freeswitch.org (brian@freeswitch.org)




-- Meet us at ClueCon! http://www.cluecon.com

[brian at freeswitch.org]

We are working to correct it. So hold on

/b

On Jun 2, 2009, at 4:53 PM, Lars Zeb wrote:

[anthony.minessale at g...]

You have a good point.

On the other hand, it's just another random day in SVN trunk. =D
Most projects don't offer SVN trunk you can play spin-the-bottle with and land on something production-ready.  But we are pretty close most of the time.

Here's my point of view:
That particular addition was a component to the core meant to be transparent.
If we did not find out the hard-way about this by adding it to trunk,
we would have found out the even-harder-way by having it imprinted in the actual release.

We try to keep the suffering to a minimum but we sometimes fall short.



On Tue, Jun 2, 2009 at 6:23 AM, David Knell <dave@3c.co.uk (dave@3c.co.uk)> wrote:

[nik.middleton at noble...]

As Anthony comments later, using SVN for updates is usually a risky business for most projects. We all have been blessed by fantastic coding to date with this project, that has lulled us into believing that using the latest snapshot will be OK. This is the first time that I’ve had problems.

I have no doubt that the DEV’s have taken this onboard, but it can sometimes be a reality check to realize that the subscribed based has grown to such a size that regression testing now becomes mandatory if the project is to move onto the next stage.

A very valid comment was made on this thread that new features should be disabled by default until thoroughly tested. It’s all part of the learning cycle. In my view the trunk needs to be updated more frequently and this should be what us mere mortals use. To often I see messages saying you’re using a 2 week old version, that bug’s been fixed.

FS, is coming to a level where code has to be managed in a more structured way, but I have now doubt this will be addressed fairly rapidly.

Regards,


From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Lars Zeb
Sent: 02 June 2009 22:54
To: freeswitch-users@lists.freeswitch.org
Subject: Re: [Freeswitch-users] Make current fails (build 13537)


Brian,

I’m probably not the only one here, but much of what I have to do to get Freeswitch going is new to me. Never installed or really worked with Linux and scripting; just a little xml. It is challenging. Freeswitch is interesting, appealing and challenging. The work your group has done is amazing. Given this, interacting with you can be intimidating.

I am experiencing the slow start with build 13532. I assume that “block all ICMP” refers to the firewall/gateway. If this is correct, why is it that I can ping the firewall from the Freeswitch box? Can you explain in more detail what it might be on my network that is blocking ICMP? All my clients and Freeswitch itself are behind a NAT firewall.

Thanks Lars

Linux fs 2.6.18-128.1.10.el5 #1 SMP Thu May 7 10:39:21 EDT 2009 i686 i686 i386 GNU/Linux

From: freeswitch-users-bounces@lists.freeswitch.org [mailto:freeswitch-users-bounces@lists.freeswitch.org] On Behalf Of Brian West
Sent: Monday, June 01, 2009 3:33 PM
To: freeswitch-users@lists.freeswitch.org
Subject: Re: [Freeswitch-users] Make current fails (build 13537)



NO its not a bad one at all. Its switch_nat_init(); in switch_core.c since your network must be eating the packets its sending out to detect if you're behind nat or not... and not getting an ICMP unreachable like it should be getting... the joys of admins that block all ICMP like idiots. ICMP has many uses... and outright blocking it is stupid. (This is my assumption cuz its what makes sense in this case)


So you're getting hit by the nice retry/timeout loop in the natpmp software we just added and possibly the upnp lib too.



So for now edit switch_core.c and comment out switch_nat_init();



I'm working my ass off to ensure that our users that do have to live in these insane nat scenarios can do so without much if any pain. Most of which uses SMB/Consumer grade routers which these two libs we added would allow us to poke holes and setup stuff and make it painless as possible.



Soon you'll have an option in switch.conf.xml to turn it off.



Please next time don't be so demanding and calling builds brain dead .. when in fact its trying to become more aware of its network config without much user input.



/b


On Jun 1, 2009, at 5:24 PM, Nik Middleton wrote:


Well I can only assume build 13537 is brain dead. Surely I shouldn’t have to edit a whole bunch of configs to get it working. FS now takes 3 minutes to start, with no indication as to what it’s looking for in the logs. That said, to date ‘make current’ has always worked well for me. Guess I was bound to hit a bad one eventually.



Still, it’s very frustrating.





Brian West

brian@freeswitch.org (brian@freeswitch.org)




-- Meet us at ClueCon! http://www.cluecon.com

[msc at freeswitch.org]

On Tue, Jun 2, 2009 at 3:14 PM, Anthony Minessale <anthony.minessale@gmail.com (anthony.minessale@gmail.com)> wrote:

[brian at freeswitch.org]

You now have -nonat and the hang on start up with the nat detection code is fixed now.

/b

On Jun 2, 2009, at 6:37 PM, Nik Middleton wrote: