EP re-registration on primary VCS after a failover

[ghislain1973]

Hello, we are running a VCS cluster with X7.1 when testing the failover our EP (C-series) do not re-register automatic to the master vcs after failover.

We need to register the EP again to the master manually.

- Some say : the registration should be automatically

- Some say : this is only auto for H.323, manual for SIP?

I would think this should be automatically

Thx,

Ghislain

ps C-series (C20/40) running 4.2 + SIP SRV records are also used

[Martin Koch]

First of all I would recommend testing with the latest software, I saw some SRV behaviors changed with TC5.

How are your SRV records? The same priority for both peers or the first one with a higher priority/weight?

If it is the same priority you would anyhow have no control where it is registered.

Within a cluster it might make more sense to have all priorities the same and see all systems as one.

Its only the master which needs to be there to configure, but you shall not matter where something is registered.

It might make more sense in a deployment where you need independent priorities and to prefer one site

over the other to run them non clustered.

[ghislain1973]

Hi Martin,

These are the SRV records:

SRV _h323ls._udp.vcsc.domain.com. 3600 IN SRV 10 0 1719 vcslux.***.com.

SRV _h323ls._udp.vcsc.domain.com. 3600 IN SRV 15 0 1719 vcsita.***.com.

SRV _h323cs._tcp.vcsc.domain.com. 3600 IN SRV 10 0 1720 vcslux.***.com.

SRV _h323cs._tcp.vcsc.domain.com. 3600 IN SRV 15 0 1720 vcsita.***.com.

SRV _sips._tcp.vcsc.domain.com. 30 IN SRV 10 0 5061 vcslux.***.com.

SRV _sips._tcp.vcsc.domain.com. 30 IN SRV 15 0 5061 vcsita.***.com.

SRV _sip._tcp.vcsc.domain.com. 30 IN SRV 10 0 5060 vcslux.***.com.

SRV _sip._tcp.vcsc.domain.com. 30 IN SRV 15 0 5060 vcsita.***.com.

The intention is that EP's registered on the primary VCS( physical located at Lux) reregister to that VCS as primary and on the other VCS as backup. Does this make sense?

[Alok Jaiswal]

Hi Ghislain,

check the cluster creation document on cisco site appendix 5.

http://www.cisco.com/en/US/docs/telepresence/infrastructure/vcs/config_guide/Cisco_VCS_Cluster_Creation_and_Maintenance_Deployment_Guide_X7-1.pdf

if the endpoint supports the DNS srv records, at the startup it will query for DNS srv and receives the DNS srv records

it registers to one of the cluster peer and keeps on using that peer untill that peer is unavailable.

in case of a failover the endpoint fall back to alternate gatekeeper keeps on using it untill this gatekeeper is un-available. In which case the endpoint will initiate again the DNS srv request and tries to register with next peer.

And i think thats the right behavior as if the endpoint will fall back to gatekeeper which it were using previously then in network fluctuation situations the endpoint will keeps on bouncing between two gatekeepers which would not be the right idea.

I hope it clarifies your doubt.

Thanks

Alok

[ghislain1973]

Hi Alok,

It makes sense to me.

Will check it further

Thx,

G

[Tomonori Taniguchi]

Please note service name  of SRV for GK registration is _H323rs. with _udp  as ptorocol and 1719 as port number.

H323ls and H323cs are for call setup.