Is it okay to run 1.1.5 harvesters with a 1.1.6 farmer?

I have a huge bunch (100+) of nodes working as plotters & harvesters at the same time, all connected to a full node as the farmer. They are currently all under 1.1.5.

However, as you can imagine, the sync is really unstable under 1.1.5 and I guess updating the full node to 1.1.6 could improve this. However it’s not practical to update all the plotter/harvester nodes at the same time as they are all running in the middle of a long job queue.

Is it okay to only update the full node to 1.1.6, leaving the harvesters on 1.1.5? Will the harvester-farmer setup work?

I have tested this on a small test network, and I find the following:

  1. The CA on the full node remains unchanged
  2. In the debug logs (INFO level) on the full node, there’s no more messages such as “new signage point to peer xxx.xx.xx.xx (harvester IP)” and “farming info from peer xxx.xx.xx.xx (harvester IP)” which used to appear every 10s
  3. However in the debug logs on the harvester nodes I can still see new messages like:
2021-06-02T18:33:52.862 harvester harvester_server        : INFO     <- new_signage_point_harvester from peer f48481d0acb3829c107c950aea903b203ac08f96c10633f282fac4b00a79713c
2021-06-02T18:33:55.205 harvester chia.harvester.harvester: INFO     1 plots were eligible for farming 03703563bb... Found 0 proofs. Time: 2.34368 s. Total 70 plots
2021-06-02T18:33:55.205 harvester harvester_server        : INFO     -> farming_info to peer f48481d0acb3829c107c950aea903b203ac08f96c10633f282fac4b00a79713c

I guess this means it’s actually working? So, is there a change in how such farming events are logged in the debug log on 1.1.6, so that I cannot see the same in the full node logs?

Any idea guys?

Since 1.1.6 you need to set LogLevel to DEBUG if you want to see the signage points going round beween your farmer and harvester. So when you set it to DEBUG on your full-node you should see if it works or not.

1 Like

Oh, that’s it…trying out now, thanks buddy!

After changing log level to DEBUG, now I can see the following on the full node:

2021-06-02T19:22:17.106 farmer farmer_server              : DEBUG    -> new_signage_point_harvester to peer f549e9646b1941ca96efb9531e4a6d336a310e4ff531646e9e29983e675b6732
2021-06-02T19:22:17.120 farmer farmer_server              : DEBUG    <- farming_info from peer f549e9646b1941ca96efb9531e4a6d336a310e4ff531646e9e29983e675b6732

So, I assume this is working?

Looks good as far as I can tell.

EDIT: I also just asked in Keybase for known issues when running different version on a many machine approach. A member of Chia Team answered it should be fine to run diff version of chia around your setup (“that should be ok”).

1 Like