Prysm: peers without a static ip and port forwarding

[ChuckNorrison]

lighthouse can read public ip from udp packets and make use of it

no chance for the prysm to get used without a static public ip or dyndns?

[ChuckNorrison]

peers should be found prior genesis to get some work done which was not expected... this genesis is a critical state how can you work with hope and luck on this? I would expect some testing peers and basic functionality prio a genesis.

[nisdas]

@ChuckNorrison This has been fixed in master , the bootstrap- node isnt baked into alpha.26. So the node will not be able to find any peers. If you can come on discord, we can get the node setup for you.

[ChuckNorrison]

Thanks for your response. This will fail for some reasons, prysm does not work for me.

I need:

• port forwardings
• IPv4 support, even IPv6 is active
• Public IP parsed from UDP packets automatically to find peers

All this will make me force to use another client

[nisdas]

@ChuckNorrison you dont need a static ip or need to portforward. Prysm is able to determine its IP from inbound packets. All you do need is a bootstrap-node. so please try with this flag

--bootstrap-node=enr:-Ku4QGQJf2bcDAwVGvbvtq3AB4KKwAvStTenY-i_QnW2ABNRRBncIU_5qR_e_um-9t3s9g-Y5ZfFATj1nhtzq6lvgc4Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpDEqCQNAAAAAv__gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQNoed9JnQh7ltcAacHEGOjwocL1BhMQbYTgaPX0kFuXtIN1ZHCCE4g
--clear-db

Please re-run the node with these flags. This should fix your issues

[ChuckNorrison]

level=error msg="Invalid address of enr:-Ku4QGQJf2bcDAwVGvbvtq3AB4KKwAvStTenY-i_QnW2ABNRRBncIU_5qR_e_um-9t3s9g-Y5ZfFATj1nhtzq6lvgc4Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpDEqCQNAAAAAv__gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQNoed9JnQh7ltcAacHEGOjwocL1BhMQbYTgaPX0kFuXtIN1ZHCCE4g provided: failed to parse multiaddr "enr:-Ku4QGQJf2bcDAwVGvbvtq3AB4KKwAvStTenY-i_QnW2ABNRRBncIU_5qR_e_um-9t3s9g-Y5ZfFATj1nhtzq6lvgc4Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpDEqCQNAAAAAv__gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQNoed9JnQh7ltcAacHEGOjwocL1BhMQbYTgaPX0kFuXtIN1ZHCCE4g": must begin with /" prefix=p2p

Already tried some forward slashes at some string positions without success

[nisdas]

sorry @ChuckNorrison , try this one.

--bootstrap-node=enr:-Ku4QGQJf2bcDAwVGvbvtq3AB4KKwAvStTenY-i_QnW2ABNRRBncIU_5qR_e_um-9t3s9g-Y5ZfFATj1nhtzq6lvgc4Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpDEqCQNAAAAAv__________gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQNoed9JnQh7ltcAacHEGOjwocL1BhMQbYTgaPX0kFuXtIN1ZHCCE4g --clear-db

[ChuckNorrison]

Ok this is working with the bootnode, sorry i was a bit mad this does not flow as i hoped and i missed a possibility to make sure everything works fine prior spadina genesis.

Further it seems the validator and slahser instances needs to always restart with the beacon instance. They will not reconnect my beacon if it got restarted. Maybe this gets solved with .27?

Thanks for your support

[nisdas]

@ChuckNorrison Understood , this was a bad on our part. We have this all fixed in the latest master, however this was not released in the current latest release (alpha.26). This is the primary reason for the current difficulties. We will be releasing alpha.27 today with all the fixes, so it should make your life a lot easier.

[ChuckNorrison]

even after .27 it needs another --force-clear-db and resync and another bunch of missing attestations got thrown cause it needs another 10 min to to recognize that some output is missing... After RPC connection success this instance hangs even in version .27 without a cleaned up database.

i would highly recommend to make that more clear and recommend that manually clear-db needs to be done.

iam starting the ./prysm.sh beacon-chain --spadina --force-clear-db, this will update and clean database. Than startup all the services...

[nisdas]

Hey @ChuckNorrison definitely understood, the issue has been patched in master so a new user will not have to force-clear-db anymore. This shouldnt be an issue as of the current moment as it only happens in certain edge cases right after genesis. Those edge cases have been resolved for now and the fix will be included in alpha.28 .

[ChuckNorrison]

this is still an issue with current beta.0. No peers since yesterday. After update it seems to work for some time. After restart the service today it can not find any peers. I was able to check for closed port 4000. After --rpc-host was set to my local ip the port is available from public.

Current startup settings:

[Service]
ExecStart=/home/nice/prysm/prysm.sh beacon-chain \
--tls-cert=/home/nice/prysm/beacon.pem \
--tls-key=/home/nice/prysm/beacon.key \
--http-web3provider=http://172.31.0.207:8545 \
--spadina \
--p2p-host-dns mydyndns.org \
--p2p-local-ip 172.31.0.206 \
--p2p-tcp-port 13001 \
--p2p-udp-port 12001 \
--accept-terms-of-use \
--verbosity debug \
--rpc-host 172.31.0.206 \
--rpc-port 4000 \
--bootstrap-node=enr:-Ku4QGQJf2bcDAwVGvbvtq3AB4KKwAvStTenY-i_QnW2ABNRRBncIU_5qR_e_um-9t3s9g-Y5ZfFATj1nhtzq6lvgc4Bh2F0dG5ldHOIAAAAAAAAAACEZXRoMpDEqCQNAAAAAv__________gmlkgnY0gmlwhBLf22SJc2VjcDI1NmsxoQNoed9Jn>

lighthouse is running fine in same environment. both are using a single geth instance from my local subnet. This gets port forwarded from my local router within another subnet

console out:

Oct 24 13:51:02 prysm prysm.sh[4774]: time="2020-10-24 13:51:02" level=info msg="Waiting for enough suitable peers before syncing" prefix=initial-sync required=3 suitable=0
Oct 24 13:51:05 prysm prysm.sh[4774]: time="2020-10-24 13:51:05" level=debug msg="Latest eth1 chain event" blockHash=0x6657733a0b0a36df4886460d579f666df6637e64023645a4f6231f2f7373d0af blockNumber=3631201 prefix=powchain
Oct 24 13:51:07 prysm prysm.sh[4774]: time="2020-10-24 13:51:07" level=info msg="Waiting for enough suitable peers before syncing" prefix=initial-sync required=3 suitable=0

[nisdas]

Hey @ChuckNorrison given that spadina and zinken are intended to be short-term testnets, peer discovery in our beta release is not backwards compatible with most spadina nodes. In beta, we updated our discovery to v5.1 which is a breaking change, which is why you see 0 peers. You can use our alpha versions to be able to run on spadina.

[ChuckNorrison]

Thanks, i was not aware of that fact. We still have some testnet flags available which do not work anymore