You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We need to review the need for channels in this structure.
All nodes in a partisan cluster should have the same channels configuration, so when using a discovery strategy e.g. DNS we get nodename and list_addrs and we append the local channels config to create the node_spec(). If a node is wrongly configured we will have two different node_spec() objects for the same nodename.
The proposal would be to remove channels from the spec and just get it from the config partisan_config:channels().
-typenode_spec() :: #{
name :=node(),
listen_addrs := [listen_addr()],
}.
This also makes the membership CRDT and the gossip payload smaller
The text was updated successfully, but these errors were encountered:
aramallo
changed the title
Review the need for the channels property in the node_spec()
Review the need for the channels property in node_spec()Jan 23, 2023
In V5 we have an updated node_spec
parallelism
is now a per channel property.We need to review the need for
channels
in this structure.All nodes in a partisan cluster should have the same channels configuration, so when using a discovery strategy e.g. DNS we get nodename and list_addrs and we append the local
channels
config to create the node_spec(). If a node is wrongly configured we will have two different node_spec() objects for the same nodename.The proposal would be to remove
channels
from the spec and just get it from the configpartisan_config:channels()
.This also makes the membership CRDT and the gossip payload smaller
The text was updated successfully, but these errors were encountered: