4

I'm currently running a node on the latest version with no config file with the following command:

./tezos-node run --rpc-addr 127.0.0.1:8732 --connections 10

I've tried with various --connections and --bootstrap-threshold values but am still encountering errors. I am getting constant peer thrashing and see:

$ p2p.maintenance: Too many connections, will kill 3
$ validator.peer(85): Worker terminated [...]
$ validator.peer(86): Worker terminated [...]
$ validator.peer(87): Worker terminated [...]
$ validator.peer(88): Worker started for ...
$ validator.peer(89): Worker started for ...
$ validator.peer(90): Worker started for ...

This all happens within a span of 1 or 2 seconds, and happens about once every minute. I am currently accepting all TCP and UDP inbound traffic on 9732 and 19732 (since it says here). My node currently runs with 2GB of RAM (AWS EC2 t2.small instance).

Also, I'm having an issue that I believe is related, in that some of my baked blocks are failing because there are no endorsements in the mempool:

No endorsements present in the mempool. Waiting until 2019-04-02T21:45:00Z (37s) for new operations.

This is followed by an injection of the block about 37s later.

When I run ./tezos-node rpc get network/connections I see about 16 connections more or less, all but one of which have "incoming": true

Any help would be appreciated.

  • I'm confused. You say you see about 16 connections, but the default limit should be significantly higher than this, so I don't understand why you see "Too many connections, will kill 3". You weren't running with --connections 10 (ish) in those logs? tezos-node config shows nothing at all under p2p.limits? Mysterious! – Tom Apr 3 at 2:08
  • Sorry, I contradicted myself a bit. I see 16 connections when I run with --connections 10, I will modify the question slightly. Regardless I see these errors no matter the number I set connections to. – user213 Apr 3 at 2:20
  • @mr z suggests you have a look at this link gist.github.com/utdrmac/2125a8514960e163837e86bc934f19b6 – Ezy Apr 9 at 2:27
2

First of all, setting connections limit to 10 is very low, you should either use the default value or use something nearer the default value which is 100. The less peer you have, the less chance you have to receive endorsements in time. Of course you have to manage a balance between a good connectivity and the processing power/memory required too handle the communications.

Concerning the message

 p2p.maintenance: Too many connections, will kill X

It is not a sign of problem at all, it is part of the peer-to-peer hygiene to accept from time to time new connections while your node already reached its connection limit. A new peer is then accepted and an old one is kicked.

It allows a rotation on connections which, among other things, prevents your node to be stuck to connections with unhealthy peers.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.