FlexFarmer v2.2.0 Released - Direct I/O, Ability to farm using self-hosted node, and more!

FlexFarmer v2.2.0 Released - Direct I/O, Ability to farm using self-hosted node, and more!

We’re glad to announce FlexFarmer v2.2.0, which includes some exciting new changes. The two biggest changes are Direct I/O integration and the ability to connect to a self-hosted node.

Direct I/O will reduce machine load, power use, and lookup times for all users.

The ability to connect to a self-hosted node means users are no longer locked to Flexpool nodes. This allows you to farm to your own node and create your own blocks. If you want to support network decentralization, you can connect to your own node, or even a friend’s via public internet without any additional configuration.

Along with that, we also added sequential (slow) plot load mode to prevent all HDDs from spinning at once, and more!

Full changelog:

v2.2.0 Release - Major upgrade with new features

- Direct I/O integration, enabled by default (not supported on Solaris and OpenBSD).
- Ability to farm using a self-hosted node, configured with `self_hosted_node_url` configuration directive.
- Added sequential plot load mode with optional per-drive delay. Please see example configuration for more info.
- Updated end user license agreement (covers all previous versions too).
- Fixed stealth networking features not working properly or triggering certain firewall systems.
- Fixed issues running large farms on macOS.
- Upgraded fsnotify for improved plot hot reload performance.
- Minor bug fixes.

IMPORTANT: With this release, we updated our FlexFarmer End User License Agreement. Do note that it covers all previous versions, not just v2.2. You can see the updated license here: https://static.flexpool.io/legal/FLEXFARMER_LICENSE.txt.

Downloads: On our Get Started page - https://www.flexpool.io/get-started/xch/flexfarmer. For docker users, the new FlexFarmer is already at flexpool/flexfarmer.

Posted a new thread as the last Flexfarmer one was crowded.


Wow some great features, maybe I gotta change to flex.
I’m a bit lazy though,:joy:


If you use a non-flexpool node, if that node has issues, will it fall back to Flexpool.io nodes?


No. Fallback is only enabled for our nodes.

1 Like

Once more question,

If I just copy my previous YML file used with previous FlexFarmer version into the newly downloaded V2.20, will the Direct I/O integration be enabled by default? I assume so, but just want to make sure.

Only the optional self-hosted node and sequential plot load mode need extra modification to the previous YML file to enable those features, correct?

Thanks for my noob questions!

To my knowledge yes.

Best place to ask FF questions is the FF channel on our discord.

1 Like

(512mb of ram)

1 Like

I have to say, getting this running on a Pi1 is pretty nuts; I can see the use case for this. Myself I’m gearing up for the digital nomad lifestyle, I still like coins (and Chia!), but won’t always have access to power or internet working out of a backpack. FF gives me the ability to jump online for a few hours when I fire up the laptop - yes it’ll be a tiny farm but I can still be in on the project (gives me something to do on my lunch breaks too).

And we’re all friends here, check it out :sunglasses:: xch1wucp2m28dqhylcyylxjs6hq49e7t0ms4jcl72r0r3a6yvlveyggs4lw6am | Flexpool.io The tiny dips on reported (and effective) space are only because the gaps in partials can be a bit long on a tiny farm - not a major concern, the payouts still average out in the end. The big dips are from the operator rebooting and screwing with the wifi, not outages. And I have the payouts shooting over to a Goby wallet so I can easily hop on and do the Web 3.0 thing I like to do.

So thanks! And I apologize for throwing rocks and trash from the open source side of the camp - it’s easy to do from an armchair, it’s hard to actually code something and contribute to the community (something I haven’t done).


Having it run on Rpi 1 is not really the same as having it being useful.

I tried to run it on RPi 3 (pre v2.2, when I still was at Flexpool), and dumped it, as lookups were getting too slow. Also, the power draw was not that much lower than my AMD Athlon 5350 APU basically making it not worthwhile to bother (that Athlon box runs at ~10-15 W when idle (basically all the time when harvesting) - miTX).

I would not mind running it on RPi 4, but basically there is no way to get it or CM4 (a better way to go about it) right now. (CM4 will give you one PCie slot with proper extension card, but the speed of that PCIe kind of sucks. Still, that is PCIe and it lets you bypass all the nonsense with USBs getting fits.)

So, if you have something like that Athlon APU, it will draw ~5W more, but will give you a fully working PCIe 3 slot that you could use for whatever you want.

1 Like


I was plotting a hard drive and was 50 plots into it… I decided to have FlexFarmer v2.2.0 also run on the plotting machine in the background and farm the plots on the hard drive as it is being plotted,

All plots after 50, (After I started FlexFarmer) are 15-20 seconds quicker to be plotted!

Who knew that FlexFarmer v2.2.0 also improves your plot times! :wink: No idea why they are quicker, but it started right after I started FlexFarmer and plot times are all 15 - 20 seconds faster thereafter.

1 Like

2.2.0 reduces I/O because its no longer in the cache buffer, no idea how that works.

I assume it improved your plotting times, as you had the chia node running before. That is basically a sample code (as stated by chia), and is rather a hog at using resources. So, just killing that node gave you those extra seconds speed up.

I didn’t have the Chia node running, just doing CLI with Mad Max.

Just checked the latest plot time and it’s back to the usual 28.6 minutes… but right after Flexfarmer started running the following 12 plots were all 28.3 minutes.

At the very least, it shows FlexFarmer is very low resource intensive as it has not affected my plot times at all… if anything right now, my times have been faster.