[OpenSIPS-Users] Dialog replication weirdness

Hi, I'm not sure whether this is a bug (OpenSIPS 3.1.2). I'm trying to do dialog clustering with the dialog module. Before clustering, I had this in the module settings: modparam("dialog", "profiles_with_value", "RX; TX") And I read the sizes in a timer route with this:

Re: [OpenSIPS-Users] Dialog replication in opensips 3.0

Is dialog replication only possible from Primary node to the backup node which is a scenario in HA ? If I am running 2 primary node then dialog sharing is possible or not ? If possible then how I will do this ? I am not getting any document which is saying how to do the setup for dialog sharing

Re: [OpenSIPS-Users] Dialog replication in opensips 3.0

Hi All , Please do help me . There is lot of confusion when I am reading the doc . For 2.4 opensips I got the document https://blog.opensips.org/2018/03/23/clustering-ongoing-calls-with-opensips-2-4/ But this is not working with 3.0 . I*s there any specific document for 3.0

[OpenSIPS-Users] Dialog replication in opensips 3.0

Hi All , I have 3 opensips Nodes running on Fullsharing clustering mode . All data in opensips memory and get replicated with each other . The cluster table looks like below . | id | cluster_id | node_id | url | state | no_ping_retries | priority | sip_addr| flags |

Re: [OpenSIPS-Users] dialog replication

Hello Liviu, Please check this diagram https://www.dropbox.com/s/od0z7wacc2yu3rr/opensips-cluster-design.pdf?dl=0 The error coming up on public inteface where no vips just round robin DNS A recoord. volga629 On Thu, Aug 16, 2018 at 5:20 AM, Liviu Chircu wrote: Hi Volga, The 2.4 dialog

Re: [OpenSIPS-Users] dialog replication

Hello Bogdan, Yes, Razvan identified issue where $fs wasn't set properly. I fixed this on script level not sure if this clean solution. https://paste.fedoraproject.org/paste/lIzEh5Q1vd4XePW6oUgC2g any insight thank you. volga629 On Thu, Aug 30, 2018 at 1:20 PM, Bogdan-Andrei Iancu wrote:

Re: [OpenSIPS-Users] dialog replication

Hi, I see the INVITE is received and sent out on the same interface : 207.210.246.38:5060, so there is no interface exchange, so no reason for a double RR. Regards, Bogdan-Andrei Iancu OpenSIPS Founder and Developer http://www.opensips-solutions.com OpenSIPS Bootcamp 2018

[OpenSIPS-Users] dialog replication

Hello Everyone, What possible cause that dialog is removing Record-Route for second interface, that cause send call to WAN instead the LAN. https://paste.fedoraproject.org/paste/XsqlLhO0CYteE3APcVkKpw route { if(!has_totag() && is_method("INVITE")) {

Re: [OpenSIPS-Users] dialog replication

Yes, I appied vips on wan and no more messages. volga629 On Tue, Aug 21, 2018 at 5:33 AM, Liviu Chircu wrote: Hi Volga, The "sip_addr" column of the opensips.clusterer table is only relevant with federated usrloc, and adding the public VIPs won't affect the cluster. Once you add the

Re: [OpenSIPS-Users] dialog replication

Hi Volga, The "sip_addr" column of the opensips.clusterer table is only relevant with federated usrloc, and adding the public VIPs won't affect the cluster.  Once you add the public VIPs, the backup will start accepting replicated dialogs, although they are being sent over the private

Re: [OpenSIPS-Users] dialog replication

Hello Liviu, If I will add 3 vips on public interfaces how to add them into database clusterer tabe, right now in sip_addr column is defined lan vips for each node. volga629 On Fri, Aug 17, 2018 at 12:27 PM, volga...@networklab.ca wrote: Thank you for reply. The point in same vm with two

Re: [OpenSIPS-Users] dialog replication

Thank you for reply. The point in same vm with two network interfaces one pubic ip and one private. volga629 On Fri, Aug 17, 2018 at 12:10 PM, Liviu Chircu wrote: Dialog replication is _built_ to work with VIPs. If you are not using VIPs, you will only be able to share dialog profiles [1]

Re: [OpenSIPS-Users] dialog replication

Dialog replication is _built_ to work with VIPs.  If you are not using VIPs, you will only be able to share dialog profiles [1] between your public front-end boxes. [1]: http://www.opensips.org/html/docs/modules/3.0.x/dialog.html#param_profile_replication_cluster Liviu Chircu OpenSIPS

Re: [OpenSIPS-Users] dialog replication

Hello Liviu, Please see the design diagram https://www.dropbox.com/s/od0z7wacc2yu3rr/opensips-cluster-design.pdf?dl=0 The error producing on public interface where are no vips only round robin DNS A record. volga629 On Thu, Aug 16, 2018 at 5:20 AM, Liviu Chircu wrote: Hi Volga, The 2.4

Re: [OpenSIPS-Users] dialog replication

Hi Volga, The 2.4 dialog clustering definitely supports "active-active" setups.  For this, you will need to use two virtual IPs, each being primary for one of the nodes, such that you avoid those "dialog created on unknown IP" errors when replicating dialogs to the backup node. Best

[OpenSIPS-Users] dialog replication

Hello Everyone, Based on this thread https://opensips.org/pipermail/users/2014-August/029635.html dialog replication is not design for active/active scanario. My question when federated no sql cluster enabled is it really need it ? tried enable and getting errors Aug 15 23:42:22 aitossbc01

Re: [OpenSIPS-Users] dialog replication

Hi Vlad, Thanks for your tip. I'll give it try. On Tuesday, 6 March 2018, 19:57, Vlad Patrascu wrote: Hi Pasan, You could have another OpenSIPS instance that exposes the service IP, in front of the 2 clustered HA nodes. Calls would first land on this box and then

Re: [OpenSIPS-Users] dialog replication

Hi Pasan, You could have another OpenSIPS instance that exposes the service IP, in front of the 2 clustered HA nodes. Calls would first land on this box and then would be dispatched to one of the 2 floating IPs. You could for example hash over the From tag or Callid in order to send to the

Re: [OpenSIPS-Users] dialog replication

Hi Vlad, Thanks for your response. I think I'll have to wait until 2.4 transaction replication come in to action if I am to go with anycast ip. I didn't understand how it would solve with 2 moveable ips. how do i make sure dialog will tie to a particular ip ? On Wednesday, 28 February

Re: [OpenSIPS-Users] dialog replication

Hi, Yes, you are correct, this is currently a limitation. But it could also work in an "active/active" sort of setup if you have 2 movable IPs and each dialog is tied to one of the IPs. Full anycast support though is coming up in OpenSIPS 2.4. Regards, Vlad Patrascu OpenSIPS Developer

Re: [OpenSIPS-Users] dialog replication

Hi Vlad, On a second thought its how anycast should work, packets could get to the closest node so basically with 2.3.3 we can't replicate transactions across mulitple opensips servers ? dialog replication  would only work for an active/passive setup doesn't it ? On Wednesday, 28

Re: [OpenSIPS-Users] dialog replication

Hi Vlad, That explains the issue then, my sip listener ip is an anycast one, and for some weired reason replies for INVITE that leaves node2 receive by node1 hence it can't create the dialog. I checked with node2 being down and recovered and calls init via node1 always received final reply

Re: [OpenSIPS-Users] dialog replication

Hi Pasan, I don't see anything in the logs or your cluster configuration which could indicate that the dialogs don't replicate. Are you sure that the problem is consistently reproducible and that the INVITE for the call in question receives a final reply? Dialogs are replicated when they get

Re: [OpenSIPS-Users] dialog replication

Hi Vlad, I have sent you the full debug logs as requested. Also clusterer_list mi command gives following output when dialog replication stopped working. node1>>opensipsctl fifo clusterer_list Cluster:: 1     Node:: 2 DB_ID=5 URL=bin:10.3.1.137:5566 Enabled=1 Link_state=Up  Next_hop=2

Re: [OpenSIPS-Users] dialog replication

Hi, Can you send the full logs for both nodes from the time that node1 restarts onwards? Also, what is the output of 'clusterer_list' mi command on the instances? Regards, Vlad Patrascu OpenSIPS Developer http://www.opensips-solutions.com On 26.02.2018 08:38, Pasan Meemaduma via Users

[OpenSIPS-Users] dialog replication

Hi Guys, Its me again :). I'm using dialog replication in opensips 2.3.3 and it appears it doesn't work after a node goes down and come back online. The recovered nodedoesn't seems to receiving dialog info via binary interface. I have node1 and node2 with dialog replication on. everything works

Re: [OpenSIPS-Users] Dialog replication problems

Hello, I backported the commit 00be97f151d254af281c4f05611c1905f51fbcde ("dialog: Fix incorrect replicated dialog ids") from git in an Opensips 2.2.2. opensipsctl fifo dlg_list show the same dialogs on both sides, but after fail-over the BYE is not routed (no problem if no fail-over). My

Re: [OpenSIPS-Users] Dialog replication problems

Thanks for the quick test! Will have it pushed upstream today. Transparent CDR generation for the replicated dialogs is currently not working. For that to work, the two nodes would need to _accurately_ share state (i.e. "who is the primary?"), otherwise double CDR/billing situations could

Re: [OpenSIPS-Users] Dialog replication problems

Hi Liviu, Yes - big difference with the patch : active server: dialog:: hash=3426:1456403041 dialog_id=14716014359137 state:: 4 user_flags:: 0 timestart:: 1478170024 datestart:: 2016-11-03 11:47:04 timeout:: 1478191625 dateout:: 2016-11-03 17:47:05 callid::

Re: [OpenSIPS-Users] Dialog replication problems

Hi, Dawid! I have looked into the problem and also managed to come up with a fix! Could you please go to your OpenSIPS 2.2 source code directory, apply the below patch, recompile the dialog module and see if it fixes the problem? git apply <(base64 -d

Re: [OpenSIPS-Users] Dialog replication problems

Anyone ? I have just upgraded to the latest 2.2 version form GIT and am still experiencing this. active server: dialog:: hash=*2297:947327686* dialog_id=9866487206598 state:: 4 user_flags:: 0 timestart:: 1478162278 datestart:: 2016-11-03 09:37:58 timeout:: 1478183878 dateout:: 2016-11-03

[OpenSIPS-Users] Dialog replication problems

Hi All, I have a reduntant OpenSIPS 2.2.1 setup with clusterer, binary interface replication and a floating IP. I am encountering a few niuances and am wondering if I am doing something wrong or if there is a bug. 1) Replicated dialog hash id is different on the standby server from the active

Re: [OpenSIPS-Users] Dialog Replication Behaviour

Hello, This bug was fixed as of https://github.com/OpenSIPS/opensips/commit/59a0360a9dbaf0c13b95ce7d038b3225ce0e4caa Please update your sources to the latest 1.11 branch, there have been several other dialog BIN replication fixes. Best Regards, Vlad Paiu OpenSIPS Developer

[OpenSIPS-Users] Dialog Replication Behaviour

Hello, I'm working on a redundant solution with Opensips version 1.11.2 using dialog replication via Binary Interface. I've noticed that the active server sends the binary create command to the backup server only when the dialog is confirmed. The problem I've found is than when a call is canceled

Re: [OpenSIPS-Users] Dialog Replication in Active Active mode

Hello Jai, Looking at your active-active configuration, I must ask: with your model, in case OSP1 goes down, how will BYEs reach OSP2 in order to terminate OSP1's ongoing calls? Your conclusions are correct. Dialog replication is designed to work with a common virtual IP between two OpenSIPS

[OpenSIPS-Users] Dialog Replication in Active Active mode

Hello, I am trying to setup two active active open-sip servers sharing the same database. Listening on IP 192.168.1.110 (OSP1) and 192.168.1.111 (OSP2) This is how my dialog configuration looks like. OSP1 bin_listen = 192.168.1.110:8080 bin_children = 4 loadmodule dialog.so modparam(dialog,