Synchronization is taking too much time

[nmnenkov]

Hello All,

I'm trying to sync conflux client version 2.4.0 (full node). The process for chunks download is very slow. Some time for 1 hour the client is downloading less than 100 chunks.

I see that the number of the peers is changed very fast. Most of the time it is going from 50 to 0 in less than 20 minutes.

I have a perfect internet connection with high bandwidth.

I have created some graphs in order to monitor the sync process.

I'm attaching hydra.toml configuration also. hydra_toml.txt

I tried different configurations for max_download_state_peers, max_handshakes, max_incoming_peers, max_outgoing_peers and max_allowed_timeout_in_observing_period. All with the same result....

Here is part of the log:
log.txt

Please advise how I can speed-up the chucks download process.

[iosh]

@nmnenkov Hi, you can use the snapshot tool to download and import snapshots to save time required for synchronization.
see: https://www.confluxdocs.com/docs/general/run-a-node/snapshot-tool

[nmnenkov]

Hi Iosh, Thanks for your answer. Yes, I know this approach. The issue is that even with the snapshot tool the client cannot sync enough fast. Now the checkpoint = FullSync height is 109240000 and each day the client cannot compensate the new epochs. This is the reason that I'm looking from some configuration optimization.

[Pana]

Where are your node located, are your node always in catch-up mode, is the un-synced epoch decreasing?

[nmnenkov]

Location is St. Louis, MO, USA.
The node is always in catch-up mode => Catch-up mode: true, latest epoch: 109282135 missing_bodies: 0
If your question is about the FullSync height - Yes, it is increasing. Last reported in the log is:
2024-11-13T19:32:08.774039537-06:00 INFO Socket IO Worker #3 cfxcore::syn - Snapshot manifest received, checkpoint = FullSync { height: 109240000...
before that it was :
2024-11-13T16:41:57.522080056-06:00 INFO Socket IO Worker #3 cfxcore::syn - Snapshot manifest received, checkpoint = FullSync { height: 109200000...

[Pana]

You can try to delete blockchain_data/net_config and restart your node to re-discover the peer node

[nmnenkov]

Hi Pana, OK. I did it (keeping the key file the same). Will see the result in the next few hours.

[nmnenkov]

Hello Pana, All,
Deleting of the blockchain_data/net_config didn't give any result. The status is the same. Even the internet provider was changed and there was no difference. Most of the time the node is downloading the chunks very slowly. Is there something that can be changed in the configuration? Are the other nodes working like this? Is it possible someone to share the log of some node that is syncing(Just to compare the behavior).

[Pana]

Our node works normal, @peilun-conflux can you help check this

[peilun-conflux]

According to the monitor graph, it looks like the chunks are only downloaded normally when the number of connected peers remain stable. You can try to change the log level to debug and check why the peers are disconnected?