weird error on federation between mainnet relays

[Omarabdul3ziz]

Versions

relay2 >> relay.02.grid.tf deployed with version 1.2.1 ninja relay >> relay.bknd1.ninja.tf deployed with version 1.2.1

Issue

two peers one connected to relay2 and the other connected to ninja relay. shows an error in relay1 which is not involved anywhere

  relays '["relay.grid.tf"]' was not reachable in time

Reproduce

using the examples from the rmb-sdk

• for server router
• for client

[muhamadazmy]

there are many parts involved here and can cause this issue to happen as follows:

• Once you start a peer. the peer updates it's information on the chain.
• The relay normally talk to the chain to fetch the twin information (including list of relays) but because this is a slow process the relay normally cash this information
  • we also have the cache warmer process which makes sure the twin information is updated in the relay cache periodically from graphql.
• This means that it can take a while before the relay detects the changes. Which causes the issue like the above.

Solution: The solution actually suggested here to implement a cache that reflects changes to the chain as they happen (with a minimal delay) so such issue is resolved immediately

[Omarabdul3ziz]

yes, verified with different twins on the three relays and they can talk to each other, so yes was mostly cached