Storage server should not conduct reachability tests or route requests when inactive
jagerman opened this issue · 3 comments
Inactive (i.e. decommissioned or deregistered) storage servers shouldn't be conducting reachability tests because those tests will always fail: the decommissioned node isn't an active SN and so doesn't have access to invoke the SN-only reachability endpoint.
This is what is causing most of the Access denied to sn.onion_req for peer warnings: they are supposed to be access denied, because the tester should know better than to try.
Actually I'm not certain that it is (only?) reachability tests, as I'm also seeing some Access denied to sn.proxy_exit, which suggests that decommissioned nodes are attempting to route (which will also fail). Rather they should return some failure immediately to the client so the client can try another node.
This seems to be fixed in the current code.