Upgrade to v15 caused the node to get de-peered

[lephleg]

Description of bug:
After upgrading to v15, the CLI stopped responding and the logs got flooded with Bootstrap stopped because there are no peers.

Steps to reproduce the issue:

1. Upgrade Docker node to v15 using the latest image tag
2. Try to execute some sort of CLI command against the daemon like /usr/bin/rai_node --wallet_list
3. Open the logs and observe the repeated Bootstrap stopped because there are no peers message

Describe the results you received:
No output at all.

Describe the results you expected:
CLI command output or an error message about the daemon.

Additional information you deem important (e.g. issue happens only occasionally):
RPC calls seems to work so far.

I also tried to delete my old config.json file and let the daemon generate a fresh one. Didn't help much.

Environment:

• Host OS: Ubuntu 16.04.4 LTS
• Node version: v15:latest
• Docker version: 18.03.0-ce
• Docker Compose version: 1.21.0, build 5920eb0

logs

[2018-08-19 23:54:33.172577]: Starting bootstrap attempt
[2018-08-19 23:54:33.172741]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172768]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172772]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172775]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172778]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172781]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172784]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172787]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172790]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172793]: Bootstrap stopped because there are no peers
[2018-08-19 23:54:33.172796]: Exiting bulk push client
[2018-08-19 23:54:33.172800]: Exiting bootstrap attempt

[argakiig]

can you stop the daemon, nuke the config, start it back up, and then provide the config.json that is created as well as what are you using to launch the docker?

[lephleg]

I did nuke the config.json and restarted the node, but didn't help much.

I managed to get it back up by deleting everything in /root/RaiBlocks/ and fast-syncing from scratch.

The fresh config.json is now the following: config.json.txt

The Docker node is spinned up by Docker Compose with the following configuration:
https://github.com/lephleg/nano-node-docker/blob/master/docker-compose.letsencrypt.yml

[argakiig]

so, it is working now? Had you been running master on the live network before this point?

[lephleg]

It seems so, yeah... But I'm not really sure if it's voting now.

Has the voting log message changed from Block XXX was confirmed to peers to Block XXX was republished to peers? Cause I can no longer see the former in my logs..

Nope, I've upgraded straight from v14.2 (the previous latest).

[argakiig]

republished does not indicate a vote, is your wallet unlocked, you have the correct seed or private key imported?

[arzarif]

Just wanted to add that I'm hitting the same issue.

I restarted the node after removing config.json and it doesn't appear to be peering with other nodes.

I'm also upgrading directly from the 14.2 release.

[lephleg]

@argakiig The wallet has no password and --wallet_decrypt_unsafe prints my seed and public address, so I suppose it's correct.

[lephleg]

OK I dont know how, but I got sparse * voting up again! 🎉

[2018-08-20 02:43:44.135225]: Block 969A1204CCE3609687E5CAF3B847392AD42141EEB304F660CE739D696AFFE8D1 was confirmed to peers

It happened after I've restarted the node again, to turn off the vote logging.

However, removing the ledger and restarting several times don't consist a recommended solution to this issue, as I can see many people dealing with it...

Edit *: The node now votes every 5 mins average, resulting in dropping uptime on nano network monitoring tools.

[BitDesert]

Had the same issue, fresh docker node on master tag wouldn't peer with the preconfigured node. Releases-v15 tag works fine.

[arzarif]

Confirming that I was also able to restore the node, but only after resyncing from scratch.

Based on MeltingIce's network status page, it appears there are several other v15 nodes that fell out of sync as soon as the latest Docker image was updated to point to the v15 image. Not sure if it has anything to do with this, though there are a couple of instances that are only showing 1 peer.

[lephleg]

Not sure if it has anything to do with this, though there are a couple of instances that are only showing 1 peer.

I can confirm that my node appeared like this on the network page, while I was having the issue described above. So, they are probably in the same state after auto-upgrading.

Now, it seems that I'm normally catching up and stay fully synced, but voting happens once in a while and not every minute (as it used to). Not sure if that was expected after the v15 release.

[rkeene]

We are going to release V15.1 to address this