We have an issue with one of our remote site. We have 2 low level relay servers in our remote site that are connecting to top level relay servers in main DC, between top and low level relay servers we have tunnel connectivity configured in such a way that if we have a traffic from top level relay to low level relay the tunnel will remain up and low level relay will continue to communicate with the top level relay. But if we not have any network traffic from top to low level relay the tunnel will go down and whole communication will break.
But now the issue is if we not have any action pushed to the low level relay (or its endpoints) for a long time the low level relay will become down and all its endpoints will be down too.
Do we have any solution here by which we can trigger some traffic from top level relay to low level relay in periodic time frame so that there is traffic from top to low level relay and hence our tunnel will remain up?
Is there a way or actions by which we can initiate periodic traffic to low level relay so that the tunnel will remain up?
HI,
We have a statement in the link that you have shared:-
"_The parent relay will try to open a socket to the child relay at port 52311. The child relay can āgrabā the socket used by the parent to communicate with it and keep it alive by sending ping messages periodically. At the same time, the child relay will start to listen on a different port such as 52312 only on its loopback address, this will be used to forward all the traffic through the socket opened by the parent that was previously grabbed."
If we apply these settings then are we sure that it will initiate continues traffic from top to low level relay?
In the statement it is mention that child will grab the socket created by parent and then keep it alive by sending ping messages periodically, if we go by this statement then we have a continues traffic from low to top relay not from top to low and we required traffic from top to low?
The Parent Relay needs to have the list of child relay hostnames or IP Addresses configured so it knows which connections to initiate.
In the article and settings thereās mention of a āpingā from the child relay; my understanding is this is not actually āICMP pingā but rather a TCP-Keepalive message on the open TCP connection, just to ensure the connection is kept open on firewalls and routers between the relays.
HI Jason,
We want to periodically send the traffic from the parent to child relay like every one hr, what will be the best settings that I can apply so that it will not impact the relay performances especially the top level relay which will be parent relay in our case.
Will the settings below be good so that top level relay will initate connection every 1hr and also it will not have much of performance impact on the top level relay.
One more query, we have 2 top level relays and two low level (in DMZ) can we apply these settings form both top level relays because we may have any one of the top level relay as primary for DMZ relays? We have manual relay selection for DMZ relays
The performance on the top-level relay should be very similar to the normal workload if the child relay were allowed to connect. If there are no new actions or site contents to publish, the only added traffic would be the TCP keepalive messages - very small messages only sent once a minute or so.
The āOpenChildIntervalSecondsā, I believe, refers to the parent attempting to re-establish a connection to the child relay if the connection fails or is broken - for example when the child relay is unavailable, how often should the parent retry. Once a connection is successfully established, the connection stays open as long as both relays and the network in between stays online.
I donāt know that itās possible to have two different parent relays initiating connections to the same child relay - itās just a case I havenāt tried.
HI Jason,
Can you please help us verify this. As we have two top level relays and 2 low level relays and both top relays are set as primary and sec (manually) in the low level relays considering if one top level relay goes down then the other relay will take over. In this setup of ours is it possibel that we can set the DMZ parent relay for both the top level relay and there child as both the DMZ relay so that persistant connection will remain open even if DMZ relay will be repoting via any one of the top level relay.
I think thatās something youāll need to test out in your environment. I donāt think weāve documented any redundancy for the parent initiating connections to different child relays, so it may not be covered in our test cases.