Closed Fofeu closed 5 years ago
I had the same behavior.
I was trying to throttle traffic a bit to please my ISP routers. I was playing with iptables
, using the hashlimit
module with the reject
target, so my own server generate unreachable
ICMP messages.
@boramalper Correct me if I'm wrong, but on https://github.com/boramalper/magnetico/blob/b0c9198f8d90eaee01c3fe0ea89752097afaf2de/cmd/magneticod/dht/mainline/protocol.go#L39 the transport is initialized with the onCongestion
function from the new Protocol
instance, which have no eventHandlers
yet, since it's done on the next line.
So either the two lines have to be switched, or the the transport should be initialized with:
p.transport = NewTransport(laddr, p.onMessage, eventHandlers.OnCongestion)
I had a crash, too but it was different. I wasn't recording logs at the time, but I'm making a SystemD unit so that I'll catch it next time around.
Shall be fixed by d7bfbc71824c577d651e3d44e647bb3ac49568ef =)
Thanks @Glandos!
I had this issue, and it does not crash anymore with the latest version.
But I still get lot of WARN WRITE CONGESTION! {"error": "operation not permitted"}
, like more than 30 per seconds
I had this issue, and it does not crash anymore with the latest version.
But I still get lot of
WARN WRITE CONGESTION! {"error": "operation not permitted"}
, like more than 30 per seconds
Hi @alcalyn,
that is likely due to not setting
iptables -I OUTPUT -t raw -p udp --sport PORT_NUMBER -j NOTRACK
iptables -I PREROUTING -t raw -p udp --dport PORT_NUMBER -j NOTRACK
as explined in magneticod readme. Some people fixed your problem with it. I still have some issue but they are likely related to some misconfiguration on my side.
Try those 2 rules (PORT_NUMBER should be 0 unless you specified some custom indexer-addr
parameter. In that case use the port you specified).
Magneticod always crashes with the same message Here is the log output before the crash