IP Failover with Sofia Recover - not recovering?!
I finally figured out how to get pacemaker to start profiles/trigger a sofia recover on moving the floating IP (but an FS crash it still can't handle) Anyway - I 1) set up odbc for the core, and mysql master-master replication, so I see calls & channels being logged in both 2) I set track calls in all the profiles So when I set it to transfer from one node to another, I'm kinda surprised I'm not getting a recovery of the calls. I do see that the DB has a hostname stored - "sip1" or "sip2" depending on where the call originated, but even if I change it manually, a sofia recover seems to ways say "no calls to recover" Am I missing something?! -Avi Marcus
I believe that hostname is the hostname of the server and set on FS startup. -Steve On 16 January 2011 00:27, Avi Marcus <avi@avimarcus.net> wrote:
I finally figured out how to get pacemaker to start profiles/trigger a sofia recover on moving the floating IP (but an FS crash it still can't handle) Anyway - I 1) set up odbc for the core, and mysql master-master replication, so I see calls & channels being logged in both 2) I set track calls in all the profiles So when I set it to transfer from one node to another, I'm kinda surprised I'm not getting a recovery of the calls. I do see that the DB has a hostname stored - "sip1" or "sip2" depending on where the call originated, but even if I change it manually, a sofia recover seems to ways say "no calls to recover" Am I missing something?!
-Avi Marcus
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
I'm supposed to have two machines with the same hostname - it's not just a variable in the xml files..? Isn't that a bit odd? And even when I modified the sql table to change the hostname in the DB and then do a sofia recover, it didn't recover. I'm not sure how to debug this, I'm just getting "no calls to recover" and I don't hear anything. -Avi On Mon, Jan 17, 2011 at 12:29 AM, Brian West <brian@freeswitch.org> wrote:
Hostnames would need to be the same on both.
/b
On Jan 16, 2011, at 3:18 AM, Steven Ayre wrote:
I believe that hostname is the hostname of the server and set on FS startup.
-Steve
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Well, partly that hostname variable predates all the sofia recover stuff so there was no reason for it originally. -Steve On 16 January 2011 22:45, Avi Marcus <avi@avimarcus.net> wrote:
I'm supposed to have two machines with the same hostname - it's not just a variable in the xml files..? Isn't that a bit odd? And even when I modified the sql table to change the hostname in the DB and then do a sofia recover, it didn't recover. I'm not sure how to debug this, I'm just getting "no calls to recover" and I don't hear anything.
-Avi
On Mon, Jan 17, 2011 at 12:29 AM, Brian West <brian@freeswitch.org> wrote:
Hostnames would need to be the same on both.
/b
On Jan 16, 2011, at 3:18 AM, Steven Ayre wrote:
I believe that hostname is the hostname of the server and set on FS startup.
-Steve
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Its so you can have multiple boxes talking to the same db and they won't clobber each others data. But in recover you have to pair them so that the secondary box can recover for the records of the primary. /b On Jan 16, 2011, at 4:55 PM, Steven Ayre wrote:
Well, partly that hostname variable predates all the sofia recover stuff so there was no reason for it originally.
-Steve
On 16 January 2011 22:45, Avi Marcus <avi@avimarcus.net> wrote: I'm supposed to have two machines with the same hostname - it's not just a variable in the xml files..? Isn't that a bit odd? And even when I modified the sql table to change the hostname in the DB and then do a sofia recover, it didn't recover. I'm not sure how to debug this, I'm just getting "no calls to recover" and I don't hear anything.
-Avi
Ah my problem was: I didn't specify the odbc on internal/external profile page, so there was no sip_recovery table. I thought specifying it for the core db was enough. -Avi On Mon, Jan 17, 2011 at 12:55 AM, Steven Ayre <steveayre@gmail.com> wrote:
Well, partly that hostname variable predates all the sofia recover stuff so there was no reason for it originally.
-Steve
On 16 January 2011 22:45, Avi Marcus <avi@avimarcus.net> wrote:
I'm supposed to have two machines with the same hostname - it's not just a variable in the xml files..? Isn't that a bit odd? And even when I modified the sql table to change the hostname in the DB and then do a sofia recover, it didn't recover. I'm not sure how to debug this, I'm just getting "no calls to recover" and I don't hear anything.
-Avi
On Mon, Jan 17, 2011 at 12:29 AM, Brian West <brian@freeswitch.org>wrote:
Hostnames would need to be the same on both.
/b
On Jan 16, 2011, at 3:18 AM, Steven Ayre wrote:
I believe that hostname is the hostname of the server and set on FS startup.
-Steve
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Well since all the sql queries will only look for stuff in the db thats for ITS host. /b On Jan 16, 2011, at 4:45 PM, Avi Marcus wrote:
I'm supposed to have two machines with the same hostname - it's not just a variable in the xml files..? Isn't that a bit odd? And even when I modified the sql table to change the hostname in the DB and then do a sofia recover, it didn't recover. I'm not sure how to debug this, I'm just getting "no calls to recover" and I don't hear anything.
-Avi
On 16 January 2011 00:27, Avi Marcus <avi@avimarcus.net> wrote:
I finally figured out how to get pacemaker to start profiles/trigger a sofia recover on moving the floating IP (but an FS crash it still can't handle)
What did you use for this part? I sent this today to get something official done, so if you're up for sharing we can both save others a lot of time: lists.freeswitch.org/pipermail/freeswitch-users/2011-March/070329.html Thanks. -- http://www.suretecsystems.com/services/openldap/ http://www.surevoip.co.uk
Sorry. I posted what I currently have here - it's a slight modification of the one from ledr in the git contrib.. something about it was broken. http://wiki.freeswitch.org/wiki/Enterprise_deployment_IP_Failover#HeartBeat.... It seems to try to restart the current one, sometimes thinks it failed.. I don't really understand it so much. I find myself doing "crm resource cleanup $fs" often... -Avi <http://wiki.freeswitch.org/wiki/Enterprise_deployment_IP_Failover#HeartBeat.2FPacemaker> On Tue, Mar 15, 2011 at 12:09 AM, Gavin Henry <gavin.henry@gmail.com> wrote:
On 16 January 2011 00:27, Avi Marcus <avi@avimarcus.net> wrote:
I finally figured out how to get pacemaker to start profiles/trigger a sofia recover on moving the floating IP (but an FS crash it still can't handle)
What did you use for this part? I sent this today to get something official done, so if you're up for sharing we can both save others a lot of time:
lists.freeswitch.org/pipermail/freeswitch-users/2011-March/070329.html
Thanks.
-- http://www.suretecsystems.com/services/openldap/ http://www.surevoip.co.uk
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
On 14 March 2011 22:20, Avi Marcus <avi@avimarcus.net> wrote:
Sorry. I posted what I currently have here - it's a slight modification of the one from ledr in the git contrib.. something about it was broken. http://wiki.freeswitch.org/wiki/Enterprise_deployment_IP_Failover#HeartBeat.... It seems to try to restart the current one, sometimes thinks it failed.. I don't really understand it so much. I find myself doing "crm resource cleanup $fs" often...
Will take a look. That's why I think we need a official one so this box is ticked! Gavin. -- http://www.suretecsystems.com/services/openldap/ http://www.surevoip.co.uk
Sorry! I knew that looked wrong! I had the original fssofia, but I've since updated it to the one I patched. Take a look again at that section of the wiki.. I think I also turned off the "stop" part because it was propagating SQL deletion of registrations to the active master... And notice the hardcoded profile names. Cleanup is appreciated :) I should get back my sql sync and maybe we can make this exactly reliable.. -Avi Marcus On Tue, Mar 15, 2011 at 12:23 AM, Gavin Henry <gavin.henry@gmail.com> wrote:
On 14 March 2011 22:20, Avi Marcus <avi@avimarcus.net> wrote:
Sorry. I posted what I currently have here - it's a slight modification of the one from ledr in the git contrib.. something about it was broken.
http://wiki.freeswitch.org/wiki/Enterprise_deployment_IP_Failover#HeartBeat....
It seems to try to restart the current one, sometimes thinks it failed.. I don't really understand it so much. I find myself doing "crm resource cleanup $fs" often...
Will take a look. That's why I think we need a official one so this box is ticked!
Gavin.
-- http://www.suretecsystems.com/services/openldap/ http://www.surevoip.co.uk
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
On 14 March 2011 23:00, Avi Marcus <avi@avimarcus.net> wrote:
Sorry! I knew that looked wrong! I had the original fssofia, but I've since updated it to the one I patched. Take a look again at that section of the wiki.. I think I also turned off the "stop" part because it was propagating SQL deletion of registrations to the active master... And notice the hardcoded profile names. Cleanup is appreciated :) I should get back my sql sync and maybe we can make this exactly reliable..
Thanks, will test this on our VMs and get a final version. -- http://www.suretecsystems.com/services/openldap/ http://www.surevoip.co.uk
participants (4)
-
Avi Marcus
-
Brian West
-
Gavin Henry
-
Steven Ayre