As I understand HA, It's better phrased as "master/distributed", not master slave. I think because a Master/Slave means the "slave" it is "dependent" upon the master, but in a master/distributed, they can be indpendent if need be (if one goes away due to an outage of some variety).
While your network is stable the HA aspect of your installation is not. Are both identical builds (32 or 64 bit)? Are either of these using real web certificates? Voicemail will not replicate over at this time. Whichever server the user voicemail is assigned is where that goes. If the voicemail goes down no user assigned to that system will be able to access voicemail. I think there's a discussion on that and it's on the roadmap t make that distributed or failover. Registrations should work. I'm not sure about AA. Anyone feel free to correct me if I am wrong or at least clarify this a little better than I seem to be able to at the moment. -- Tony >>> Chris St Denis <[email protected]> 06/11/09 6:37 PM >>> Scott Lawrence wrote: > On Thu, 2009-06-11 at 14:47 -0700, Chris St Denis wrote: > >> Scott Lawrence wrote: >> >>> On Thu, 2009-06-11 at 13:47 -0700, Chris St Denis wrote: >>> >>> >>>> Scott Lawrence wrote: >>>> >>>> >>>>> On Thu, 2009-06-11 at 11:15 -0700, Chris St Denis wrote: >>>>> >>>>> >>>>> >>>>>> (resending this because it hasn't gotten a reply, and I see it tagged as >>>>>> spam so it may have gotten filtered for most people) >>>>>> >>>>>> I've installed 2 sipx 4.0 servers. >>>>>> pbx1.smartt.com - master server, fully configured . >>>>>> pbx3.smartt.com - slave server, joined with sipxecs-setup >>>>>> >>>>>> However, Job status shows replication is failing. Seems to be SSL >>>>>> errors, but the sipxecs-setup seems to copy over the certs and keystore >>>>>> so I don't know why it would have SSL issues. >>>>>> >>>>>> Is this a bug, or did I miss some important step? I haven't been able to >>>>>> find any documentation on sipx4 replication (all I can find it 3.x which >>>>>> seems pretty different) so I'm just going by the info in the sidebar on >>>>>> the add a server page on the master. >>>>>> >>>>>> >>>>>> Here are some logs from the master of the failures: >>>>>> http://pastebin.ca/1453932 >>>>>> >>>>>> >>>>>> >>>>> What specific build are you using? >>>>> >>>>> >>>> Server is running sipXconfig (4.0.0-015321) >>>> >>>> >>> It's very likely that your problem is fixed in the 4.0.1 builds. You >>> can either upgrade to the most recent one in the Developer Builds area >>> (be careful - you want 4.0.1, not 4.1.0), or wait a bit until we copy >>> it to the stable area (almost certainly tomorrow, by the way) >>> >> Thanks. I'll try a yum update tomorrow and see what happens. >> >> Hopefully that will fix my other problems too -- I've been getting >> weird database errors in the log (despite the web interface working >> fine so it's not a complete database failure) >> ==> sipxcallresolver.log <== >> "2009-06-11T11:15:16.982670 ":ERR:Loss of connection to >> database - retrying to connect after sleep >> >> ==> sipxstunnel.log <== >> 2009.06.11 11:15:16 LOG5[4558:3086637968]: Postgres-1 >> connected from 127.0.0.1:47880 >> 2009.06.11 11:15:16 LOG3[4558:3086637968]: connect_wait: >> getsockopt: Connection refused (111) >> 2009.06.11 11:15:16 LOG5[4558:3086637968]: Connection reset: 0 >> bytes sent to SSL, 0 bytes sent to socket >> And the ability of phones to register has stopped working as well >> sometime after I added the replication. >> [Phone to pbx] >> REGISTER sip:pbx1.smartt.com SIP/2.0. >> >> >> [pbx to phone] >> SIP/2.0 100 Trying. >> >> >> [pbx to phone] >> SIP/2.0 408 Request timeout. >> >> Not sure if these are related to the replication problems, or even to >> eachother. >> > > They are related. > > Basically, my advise is that you get the replication working reliably > before you do _anything_ else. Don't configure phones, don't try to use > anything. I know that sounds drastic, but unless you have good reliable > distributed management, many features are going to be very flakey. It's > not true that a stable system can't tolerate network outages between > systems - it can, but you can't get a system to be stable in the first > place without the replication working. Our network is stable -- The 2 servers are in the same data center and in the same subnet. The replication is wanted mainly for high availability/failover in case of hardware failure, and to minimize the disruption if the main server has to be taken down for hardware changes. I have already created all the phones and users, so I really hope it will be able to replicate over the data. It looks like the master generates the xml files then simply copies them over to the slave and restart is so it seems like it should be fine to add/remove replication slaves at any time. Tho there is probably more complexity than I see. The information I have read seems to be a little unclear about what can be replicated. The page with server roles suggests only sip routing can be replicated, and some things I have read suggest that even registrations aren't replicated/slave can't handle registrations. Weather voicemail can be replicated also seems to be unclear. If the main server fails, does voicemail work fully, stop working entirely, or accept new messages, but not replicate them? Will the auto attendant work on the backup server? _______________________________________________ sipx-dev mailing list [email protected] List Archive: http://list.sipfoundry.org/archive/sipx-dev Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-dev sipXecs IP PBX -- http://www.sipfoundry.org/
