Allow visor to start in case of address resolver or tp disc failure #402
Conversation
…ed in terms of not-yet-ready client
…nto feature/allow-visor-to-start-without-tp-disc-and-addr-resolver
Darkren
changed the title
Darkren
requested review from
nkryuchkov and
evanlinjin
and removed request for
nkryuchkov
| addressResolver = ar | ||
| if conf.NetworkConfigs.STCPR != nil || conf.NetworkConfigs.STCPH != nil { | ||
| var ( | ||
| addressResolver arclient.APIClient |
There was a problem hiding this comment.
When I first made the snet library, I intended it to be just a wrapper for all the network types. I think it would be cleaner if we didn't have any reconnecting logic here at all. Would we be able to implement the reconnect logic within arclient.httpClient?
There was a problem hiding this comment.
@evanlinjin @Darkren Would you mind if I work on the reconnection logic of arclient.httpClient? Its code is familiar to me and I have some ideas for reconnection logic. Moreover, I'm going to add a UDP connection in #400. In my opinion, it would make more sense to implement reconnection logic after UDP implementation is finished.
There was a problem hiding this comment.
@nkryuchkov I'll leave a TODO comment then
There was a problem hiding this comment.
Works fine. Can be merged if @evanlinjin is happy to fix the rest in another issue tackled by nikita.
…hout-tp-disc-and-addr-resolver Allow visor to start in case of address resolver or tp disc failure Former-commit-id: e4df449
Did you run
make format && make check?Yes
Fixes #401
Changes:
Tested partially, except for the transport manager initialization and already-known transports. We need to have
stcpr/stcphtransports working to test this. Also code may look clumsy in some places, but this should get sorted out as soon as we deal with the refactoring of network part which is proposed by @nkryuchkov