If that was for me, thanks, and I know, but I’m happy running it, bought extra ram for this box so it’s not an issue.
I knew 16 GB would be to close for comfort so I have 32.
Don’t you farm with your full node?
@drhicom
Yes.
I have two identical boxes (hardware wise).
Box #1 runs everything, as well as plots (all via batch scripts).
Box #2 only plots (via batch scripts) – starts no chia processes, other than running "chia plots create blah blah blah).
It is my understanding that one box has to be a full node, running everything (but does not necessarily need to do plotting), and additional boxes can run farmers independently, and if configured correctly, will send their farming results to the one-and-only-single-harvester that is on the full node.
That is why I do not understand whether or not a box, that does only farming, needs to be synced?
I thought that having more than one box syncing behind the same router (behind the same internet gateway) will create conflicts as to where incoming challenges should be forwarded to?
That is my understanding. But I am not confident of the above, which is why I inquired about syncing on a box that only does farming.
It’s not advised, but many have done it without issue.
Ppl even run different nodes behind the same ip without issue.
That’s not quite right.
A box that is only plotting does not even need to be network connected.
Your full node could run farmer, harvester, wallet too then additional boxes could run just harvester which is easy.
Indeed it’s likely due to the dust storm yesterday and the Python node.
I assume that is the pool code chia provided for operators. Are you going to get it improved (better H/W, rewriting the offending code)?
During the last week dust storm Flex pool did rather poorly comparing to other pools ( https://twitter.com/XCHcentral/status/1473155689396637697). Space pool didn’t flinch. Some smaller pools were less affected. So, that implies that it is not really size related, but rather poorly operated pools.
Looks like yesterday was no better. It would be good, if you could provide some timeline about fixing it.
Actually, XCHcentral has update about this last storm. Flex is at the very bottom this time. Some other pools improved a lot since last week storm (https://twitter.com/XCHcentral/status/1475832805984665600).
No we don’t run the reference pool code but we’re stuck using the same Python node everyone else uses. We’ve added some secret sauce but a full rewrite is needed.
Space only requires 300 partials be sent a day which means their test rig could have submitted a partial before and after the storm without any issues being noticed. Of course it could also be that space wasn’t affected and if so that’s great in the end who knows why they did so well.
That being said Alex says we should have been at 100% but were in the middle of an upgrade when it happened.
And to point out, we’re #2 on the pool profitability comparison between the 4 major non Chinese pools on that site. Dust storm doesn’t seem to have affected our profitability, infact we’re kicking ass.
Considering that most of Flex farms were getting about 5%-10% stales, if your code would not be affected, you would be kicking even more asses. Hopefully, that will happen during the next storm.
By the way, I have also browsed few top Space farmers, and have seen several that looks like were not affected. That was not the case when browsing Flex top farmers.
Thank you for that link to xchcentral reporting page. Actually, when you compare Space to Flex, Flex had much better results reported for the last 7 days, but Space came really close during the last 24 hours, what further supports the notion that Flex didn’t do that good during that storm (~5% advantage for 7 days, vs. ~0.5% for the last 24h - considering that this storm lasted about 8 hours, it had to be a big dip during that time on Flex side).
Space Ftw ( for the win ) !
I’ve had 17 filters pushed to 5 to 10s, and thats all.
Won’t affect a win at all.
And the + , I’m not centralising by using flex.
U r the OG AFAIK 
Maybe one day I will understand 100% of what you write
First, by joining Space pool, you support creating a pool monopoly that drives those smaller pools out.
Second, there is Flex pool and Flexfarmer. The first is not centralizing, as that is just a straight pool.
Third, why would you penalize someone that does development that addresses problems with the main code, that is basically neglected.
Fourth, Flex promised to have a full node in a couple of months or so. That will be 1:1 replacement for the original code, but supporting more H/W. I don’t see such development coming from any other pool or fork.
So, as a developer, I give them credit for doing that work.
1 Like
The ftw explanation was just for you 
.
I wasn’t aware flex had a non centralising solution, else I’d have considered using it, guess I’ll give it a look.
Sure they develop, but when I look for a pool, I only look for a pool, nothing more.
Thanks for making me think though.
Enjoy the rest of the hols, hope you had a good one.
1 Like
I know, that’s why I wrote that “U r …”
Also, all the best to you!
1 Like
Us pool ops have a private chat. It has been pointed out that some pools may use different settings. One example is that we may see shares over 25 seconds old as stale while others may be up to 60 seconds. Would need someone to test out pools to know. So we and other pools may report more stales because we’re more strict, who knows. 100% versus 95% during the storm doesn’t mean more money because a partial thats 50 seconds late thats still accepted by the pool for income purposes would likely not be accepted by the network for block finding purposes. So everyone appears as 100% but you get the same income as the 95% pool. I have no evidence this is so though (and no evidence against) so my point is only that its not a 1-1 comparison like you would have if you just had a rig on both pools and compared profits (which we do have and we’re doing quite well).
I guess, my point is that you provided that link to xchcentral. It is clearly explained on that page what they do (have the same number of plots with all those pools), and what they measure (payments for those plots at the same intervals). There is nothing convoluted about that, at least for me, as my limited observations agree with that report.
The best what I can say about your reply is that you are trying to muddy the water, what I really don’t care.
The obvious question for me is whether after analyzing that report, you will try to fix issues on Flex side (as other pools apparently did). Maybe you want Alex to review that page once more.
As I stated, I am Flex pool member, so not really your enemy.
Thats correct but when the plots are finished being made you have to copy them somewhere and sneaker-net won’t cut it these days…
1 Like
Not a the Raspberry Pi pricepoint.
Intel NUC’s are also small, relatively low power and well connected.
If you can still get for instance:
Intel Next Unit of Computing Kit NUC7PJYHN - Barebone
mini-PC - 1 x Pentium Silver J5005 1.5 GHz 4C - RAM 0 GB - UHD Graphics 605 - GigE - WLAN: 802.11abgnac, Bluetooth 5.0
and put in 8GB of RAM and any decent 2.5" SSD you’ll be fine.
It has 3 USB3 ports, but you can always use the RPi as a remote harvester for expansion of course…
New NUC’s are even better of course but so expensive right now and even then hard to get.
This one makes a nice GP PC or tiny server if you’d want to reuse it some time.
I’m running a MMX node on it now and with it’s GPU it’s doing fine.
On any system you choose, do lower the target_peer_count in config.yaml or disable upnp and port forwarding in your router, that will limit the number of (outbound) peers to default 8.
Currently I’m using a Intel J4125 with 16 GB of RAM.
The NUC costed 200 euros and the extra DIMM you can buy for 65 euros.
The NUC had a 8 GB DIMM. You will only need to replace the RAM if you want for farm several forks at a time. In that case, use Machinaris https://github.com/guydavis/machinaris
I recommend a beelink or a minisforum mini pc, since you can replace the RAM, add a 2.5" drive and has a reset BIOS button if you mess up the BIOS configuration.
To sum it up, for chia farming and only 1 or 2 forks, 8 GB of RAM is enough. I changed the CPU TDP to 6 Watts.
I have accepted that this is probably gonna be the new normal until some kind of mandatory fee system is put into place. Doesn’t really bother me as I’ve never been affected by ANY of the dust storms. Only thing the dust storms have done for me is cause me to win more lol. I do think the team should put a mandatory fee in place though. Would probably stop this kind of stuff quickly. You don’t see anybody spamming ETH’s network do you!?