-
Notifications
You must be signed in to change notification settings - Fork 265
Significant reward drop. Affecting mostly DIY miners. Seems peerbook related #1339
Comments
This is not new and there's already a github issue for this, but it has been closed: #1238 |
This is not just a small fraction. This a complete drop in rewards as if these miners were shut off the network for a seemingly specific group users. When this was the same issue, you would expect that the drop in rewards wouldn't last for over 10 days now with consistent 0 rewards. It is not the case of "failed to send witness, max retry". |
Don't give too much attention to the title, check the content. It's the same issue where witness receipts are delivered but there's no poc receipt in the blockchain. 50% of my miner delivered witnesses (around 46%-47% to be exact) end up with no poc receipt in the blockchain. And this has been happening since the start, even when my sensecap was stock. |
I have onboarded 4 new units several days ago. They worked for two days and are now completely flatlined as well. So far the only response I have seen is one of the team members saying that they don't see this in their data and the network is running(and thus the issue doesn't exist?). It's disturbing. |
Regarding potential peerbook issues, the fact that witnesses are delivered to the challenger ("successfully sent witness" in the log) seems to indicate that the DIY miners peerbook it Okay. That log message is written when a connection is established with the challenger but technically before the witness data is actually sent on the connection. Nonetheless, if there was a peerbook issue getting the challengers p2p listen address, this message would not be seen as a connection to the challenger would not have been established. This may indicate that the witness is being dropped somewhere else along the way - either in the transmission over the connection or during validation by the challenging hotspot. On the beaconing drop off, the lack of receiving challenges could still indicate a peer book issue. The miners address may not be appearing in challenger miners' peerbooks or unable to be resolved via ARP. This is evident by the lack of log messages showing that p2p messages are being received and decrypted by the onion server for beaconing. |
I have the same issue, with a complete wallet. These miners are professionally installed in radio towers, and were all earning 1HNT+ per 24 hours. From the 1st of January this has been decreasing, with 0,25HNT on 1-1 and now 0.08HNT per miner. This is a screenshot of the complete wallet, with 67 miners in it: 30-12: 47HNT The drop on the 26th and 27th is expected as there were issues with the internet connection. I've checked, and the connection is stable now. There is 1 miner in this wallet that has "custom" firmware: the packet forwarder and miner run as a docker container on a linux arm64 OS. These containers are the open-sourced containers from Helium without any changes. When I check the logs for this miner I see many messages that the witness has successfully been sent to the challenger, but almost none of these gets rewarded. Please see the following screenshots. In the miner logs there are clearly multiple witnesses sent in just an hour. However, I only got rewarded for 1(!) witness for the whole day. Here are some screenshots of other hotspots in that same wallet: |
I have the same problems,can we talk about more details,can you give me ur emails? |
#1339 (comment) |
Same like this issue, please help to anlyze. |
@JJsilvera1 this is exactly the same issue i am experiencing right now with my sensecap. From nearly 0.5hnt a day dropped to 0.02 |
It was noticed by several people that the few witnesses that are registered on the blockchain come from challengers from makers FreedomFi, Syncrobit and PantherX. Are they late with updating the latest image? What mechanism has changed that other brands like sensecap and bobcat never seem to be able to complete a PoC submission from the witness? |
All of my hotspots also flatlined, 6 of them at the same time. Decrease in rewards started in January 3rd but flatlined in 5th. All hotspots lost %90 earnings. All was above average. |
my hotspots also only get reward from challenges |
If this is true. Why not the helium technical team transfers the project to another team who can hold all the situations. Instead of always putting us to wait and wait. |
I have the same issue on my bobcat since 29 Dec. Around me everybody is earning 0.5hnt a day while mine is at 0.05...almost no witness and beacon 0 0 B |
Ditto. I notice the logs are sending the receipts, sometimes it needs to try 2-4 times . Is this an issue on the Erlang and MAC address? Is MAC or HARDWARE address some how used in the the Erlang to obtain a connection / accept receipts with peers? Could p2p peers with a ports with 2154 or any other random port not opened by the container be an issue? Nebra miners have a upnp container, Should one be implemented by all hotspots, or any further ports be open either outbound inbound on the container, or on the network environment? The receipt gets accepted, but I have not seen all the witnessing rewards on receipts. I understand not all witnesses get put into the PoC rewards if its over X amount of witnesses I think 10 now .. and that is " randomly selected" are the hotspots that dont "win" the PoC reward if total witnesses are over X amount not get their " receipts reported"? and/or are " invalid witness sent and then discarded by the end peer due to a " flagged" invalid status of some sort? |
One of my OG miners is only able to see beacons from other OG miners. It no longer is able to get rewards if it sees bobcat or rak or any other kind of manufacturer. |
Could this be a lora_hal issue? incompatible? Is the sx130x_hal Packet Forwarder still the defacto? or has it changed to a newer packet_forwarding service that isnt the Semtech flavor or something different? The packet_forwarder from lora-net sx1302_hal is what im running which is the Semtech UDP packet forwarder. Is it possible that something in the msg header that is conflicting with the recipient of our PoC receipt, like if the msg string in the MAC or euid could be formatted differently causing this? ** also I notice in the logs whenever a client connects to the miner to send a message, it isnt using port 1680, its using the second port number in the sys.config ..
I noticed that the 31341 port is the one being used. Should our container reflect this or our networks? |
I have the same issue. What’s weird is since 4 hours ago it has picked back up and I’m seeing more activity. |
Did you do something? I already power cycled, reset my blocks and start from new snapshot + turbosync (sensecap). But still no beacons for 7 day's now. I'm not relayed, and sending challenges, but these never go over 2 responses. mostly 0. Thanks! |
I did a reboot about 4 hours ago. I did also reboot yesterday and the day before and those did nothing. My miner is always fully synched and has good internet connectivity. |
Thanks to point this out. Going to try the next workaround, put miner in relay status by closing port. Seems to be counter productive. But anything is better than only creating challenges... (and maybe wait on an official statement from the Helium team). |
Already tried that. It doesn't help either. |
Does anyone know where these seed node changes would show up? Is there a relevant repo? Or blockchain transaction type? We should be able to find what was changed, and who was responsible. |
@dogweather As far as I know, the code running on the seed nodes is built from this repository. See tag 'seed0.0.45'. |
After extensive debugging and reverse engineering, the issue was found. If you want to know whether you are affected, I created a github repository with a tool that returns true (if you're banned) or false (if you're fine): https://github.com/FezzFest/check-helium-bannedlist |
This might help solve the issue. https://github.com/FezzFest/check-helium-bannedlist .created byFezzFest |
YES! Found my hotspot in the list... (eager-watermelon-wombat) I do not know why, I have a regular Sensecap M1 with an 5.8dbi omni on my roof. Normally earning only a small amount, 0.1 to 0.2 HNT per day. At least I don't need to keep debugging all things I can think off, just waiting on the new 2022.01.11 firwmare. I hope if the list comes back, I'm not on it anymore... |
Ok so let me get this straight: So let me line things up: Heads up to the hard work of the community members who did the work over several days to find and prove this as Helium obviously decided to not tell anyone till they were forced to break their silence today. For me that's a very weak stance but I let you all judge for yourself... Check below if yours on the list or not The discussion on Reddit: https://www.reddit.com/r/HeliumNetwork/comments/rywhtm/3000_hotspots_flatlined_since_dec_27th/ |
Ughh this seems all sorts of messed up. Cus if they are black listing people based on “statistics” if I’m 1 wallet with 1 hotspot on it. And I get black listed, and the person with 20 hotspots on 1 account doesn’t … “statistically” speaking the person with 20 hot spots on 1 account is more then likely the person “gaming” if they are . So the statistics of the 20 hotspots vs my 1 makes me the “gamer” ?? If all the RSSI data and all the SNR data is being quantified from “false data” then the person with accurate data is looked as the “gamer” ..? I'm not on that list... So That doesn't explain very much for me. Unless the people around me are on that list... or the peers / challangers are on that list, that could indirectly affect hotspots not on that list. |
@FezzFest Can you show us the source of your blocklist data? I looked through your commits and source code but couldn't find info about how you gathered the data. Your work looks excellent, but I'd like to verify it. |
Wow, those are some unsavory coding practices: all those dependencies are pinned to particular commits. This makes the changes completely opaque and difficult to follow. |
@dogweather The code was extracted from the miner_poc_handler.beam file in the docker image hotspot vendors pull from https://quay.io/repository/team-helium/miner. You can disassemble the file with the following erlang oneliner: The blacklist was only in the compiled images, it was never published in any of the github repositories. |
Any news about how to appeal? My whole fleet is blacklisted and my hosts and I are getting nervous about the way Helium (doesn't) handles this .... |
They are going to remove the list in the coming release. When voting is done they are restarting this list again. It is currently unknown who will stay on the list and who won't. There is no way to appeal. Helium thinks they know 100% for sure who to ban and who not. |
How and where is the voting take place? Or is that only for validators? |
heliumvote.com |
I have a problem that I haven't been able to solve for about a month. I'm using the Sensecap M1 and suddenly I've had a drop in the number of witnesses. Everything is normal, the device is not transferred, the synchronization is ok. Also, my device is not blacklisted. I'm tired of looking for a solution. |
I know this isnt available through the normal or what is "standard" setup using the ARM64 image, I had syncing issues that the API didn't update my syncing status for 2 weeks. I ran the miner on AMD64 image while still running the ARM64 miner like normal, and it updated the syncing within an hour of doing that. Ive written several posts and have documented everything somewhere on this Issues board, I don't work in IT nor am I associated with anyone or anything, My non " expert " hypothesis is either the manufactures are using their own compiled software which conflicts with other platforms. But this is not manufacturer specific as all my tests have been using Helium Official REPO and Helium Official Documentation. How I got the 2 different versions to work at the same time is " Not official* I ran that as a test and that's what I concluded. I dont think anyone else on this planet has done this, Or maybe they have but Im the only one that seems to be documenting it so that the Dev Teams or someone that knows how to code, can use it to help better understand whatever issue they are having between dependencies. CLEARLY the network is compromised. I suggest and propose, a STANDARDIZED dependency Library, The main repo on Helium miner is, STANDARD. but the dependencies are not. They are compiled from diffrent dependencies and/or versions of Erlang or LibP2p whatever it is, Stems from that. They*(ARM64 vs AMD64) If you want to get into conspiracies do that on your own time , you got the right to. But Some may believe it to be " Gaming" between manufacturers and or " users " I don't believe this but You never know in crypto " game " that could be considered a way of " Gaming the network" But note, Validators need to run in or on an AWS server, Which uses AMD64 firmware image. AMD64 firmware image requires a AVX capable core, A AVX Iimage * can not run on some ARM64 processors. BUT the ARM64 firmware image can run on any AWS system. I Even think its more economical to run ARM64 images on AWS paid servers vs the AMD64 version, It could be vice versa but I know that to be true. Basiclly the way to fix it from a non " expert " point of view. Is plain and simple, not complicated. Use 1 image that works across all platforms, That image exists. The ARM64 version. If they just stick to 1 official image, Then that would solve this issue. Does anyone listen to me, My wife doesn't, So Im sure Helium wont either. |
Nebras have the same issue! |
Im not the original mind behind the helium miner master folder but i might be able to aid a bit
The miner master folder is a compilation of settings for the most common devices breaking down each the code attached can be both helpful and confusing. I would use it as a template to find out a little more behin the scenes and what the standard is for flow for a machine which is desired
My helium florida archive is just this it contains miner/master but its contents have been used with balena os and helium console. The platforms tested on both raspberry pi 4 aarch and pi 4 arm64 which are both available in the m1 build by seeed
First off it might not be u. It may infact be a new obstcle or structure in line of site. To test the theory you could use helium rf to find out what your line of sight os and if there is any natural or structural blockages.
Secondly. I would consider running th m1 on a second antenna linked by t connection.
M1 antenna up wards and a lead out to antenna two. Go for 8-10 dbi Also the m1 lorawan gateway chip support 12.8 dbi without antenna. You could boot up then plug up your antenna after boot.
Then get a slight boost off if you dont have a option for a second antenna.
Reflections space is a factor. Im sure you know all this. Outdoor mounting is best. Omni directiinal always on antenna 1 and directional if you know your target direction
also another soft remedy is rewards scale and virtual placement with hotspotty try for seven hundred or better connection score in your strategy.
Weed out any connected network below .60
Hipe this helps a little
…Sent from my iPhone
On Jan 19, 2022, at 4:43 PM, serbyxp ***@***.***> wrote:
I know this isnt available through the normal or what is "standard" setup using the ARM64 image, I had syncing issues that the API didn't update my syncing status for 2 weeks. I ran the miner on AMD64 image while still running the ARM64 miner like normal, and it updated the syncing within an hour of doing that. Ive written several posts and have documented everything somewhere on this Issues board, I don't work in IT nor am I associated with anyone or anything,
My non " expert " hypothesis is either the manufactures are using their own compiled software which conflicts with other platforms.
But this is not manufacturer specific as all my tests have been using Helium Official REPO and Helium Official Documentation.
How I got the 2 different versions to work at the same time is " Not official* I ran that as a test and that's what I concluded. I dont think anyone else on this planet has done this, Or maybe they have but Im the only one that seems to be documenting it so that the Dev Teams or someone that knows how to code, can use it to help better understand whatever issue they are having between dependencies.
I suggest and propose, a STANDARDIZED dependency Library, The main repo on Helium miner is, STANDARD. but the dependencies are not. They are compiled from diffrent dependencies and/or versions of Erlang or LibP2p whatever it is, Stems from that. They*(ARM64 vs AMD64)
—
Reply to this email directly, view it on GitHub<#1339 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AICUODUZCKKZIP3WYS2P34TUW4WBRANCNFSM5LMNFKMQ>.
Triage notifications on the go with GitHub Mobile for iOS<https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android<https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>.
You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>
|
I've been looking into why my hotspot (good setup, beacons never get more than 10 witnesses) and my neighbouring hex hotspot (good setup, never more than 10 witnesses) could fail to witness each other so often. In looking through the last 24 hours of my HS logs, I found my HS had witnessed 24 times Out of htose 24:
I turned on a debug level of logging to investigate the failures to connect to the Challenger and found that my hotpsot would always get failed to dial challenger "/p2p/...": not_found Until this happens: And then the next connection works. When my HS times-out from too many retries, it never has that log from libp2p_multistream_client Here is a typical witness:
|
I’m in Florida and I’m a dense area , but I have been beaconing etc, but most come back with the full 14, some come back with less then 10 and I’ve had 0 as well , but I noticed there is a new error that I’m getting some yamux dial error I haven’t dug into it more but, I also have a lot of errors for multi stream dial, which from my light digging seems to be when I challange or when I beacon, after all the witness receipts come in I get them I receive them but then when I go to send them one of those errors occurs . Not sure what or why. But they fix one thing and a new thing pops up band aides for everything on the network .. I read somewhere that they keep doing a bunch of stuff without putting it up to a vote , I’m already beginning to believe some of the conspiracy theories out there. It almost seems like it’s being done systematically, or on purpose. But whatever. Just going to wait until this thing breaks even and re sell it. Or just forget about it , wasting to much time on something that never gets fixed. heh also I’ve gotten tons of notifications from Amazon and other apps / services that I use that my passwords (which are all diffrent ) are trying to be logged in from a bunch of Chinese ip addresses. So I mean it’s not heliums fault, but by having to disable a bunch of my firewall settings etc having the ports open on my modem is compromising my home network. Never had that issue until now.
I also have this issue. Sometimes not the 10 hotspot minimum but that it just seems like PoC is broken. I read somewhere also that theirs a new PoC calculation of some sort, not sure if that’s official yet or in the works. Had something to do with rewards and witnesses , what I mean by witnesses is what has been posted about witness list, if you have x amount of witnesses in your list, the calculation does some weird thing that I didn’t understand about how much rewards you can get from beaconing / witnessing . I don’t know 🤷♀️ it was strange didn’t make much sense to me. But it’s almost like a CAP per hotspot in a sense . I know the beacons are like capped at 3-4 per day now. Also their was the blockchain outage that messed things up the other day. But I have friends who’s hotspots haven’t beaconed in weeks or months… they witness a lot but even with their 100s of witness rewards they are still doing worst then anyone that gets the 3-4 beacons per day. Which is less then 2.00$ per 24 hours , if you beacon your Atleast gaurenteed 1.50$ per day but then if your getting a max of 10 or less your losing out on the % from the difference of 14 . So I beaconed with 0 witnesses because the challanger didn’t receive any of the receipts, then that’s one beacon that you get screwed out of. I mentioned it before as a suggestion , that they shouldn’t let relayed miners participate in challanges. It should be a metric of some sort for creating challanges. Because it messes up everyone else that’s running their hotspots correctly . *The multi stream thing you can see in the console log, but you got to filter or find all the errors or warnings etc… theirs also another error that occurred which is an IDENDiFY/numbers fails to handshake that’s strange also. And it closes the stream not sure if you’ve got that going on to another thing I have noticed based on my light dig, was that all the hotspots in my area beacon at around the same time. So it seems like the network / beacon algo is doing something that’s geolocation specific, wether that’s always been that way or not, or is part of the anti gaming metrics , but it seems like when I beacon then I get witnesses and after a certain amount of time then the network goes dead around me. I don’t know maybe that was just on that day that I looked at which happened to be when the blockchain outage occurred . |
@serbyxp I understand your frustration! The network obviously has a lot of fundamental problems, and because we can't get the full picture we feel left in the dark and vulnerable to all the gamers that are taking advantage of it. I get the impression the core developers are well aware of the problems with libp2p and the network in general. I don't know yet if they are up to the task of fixing it... but let's give them a chance. I am trying to capture the failed transactions in detail so that it might help them debug their code. |
How long has this issue been going on? Why can't anyone spend some time and affect it as it is not specific to one miner, no blacklist involved - it's a great many people and it is completely wrecking the network. Yes, there's lots of other issues to fix... I get that. But we as a miner community are very frustrated and the issue is out of our hands. Please. Please. Please. PLEASE fix this libp2p "failed to dial challenger" witness issue. Thank you! |
From my understanding the network is moving to gRPC so libp2p is going to be depreciated, that's probably why they aren't to focused on the issue, if they are focused on the gRPC release? not sure when that is supposed to come out , but that's when the network shifts from full miners to light hotspots. So... I mean kinda just spinning our wheels trying to debug things if they aren't even going to implement anything new to libp2p. |
Does anyone know a way to change the retry time? Here are my logs. And as you can see, it try to dial the challenger and fails. The miner then wait some time and try again and it is successful then. I want to change the time between because even after the successful message the witness still dont reach the end and i dont get a reward. I suspect that it took to much time.
|
All: This issue was most likely due to the hidden denylist that was in place at the time. I think we can close it now, as the denylist itself is public and almost certainly explains what was observed. |
Since the Dec 29th update a growing group of users is reporting a significant reward drop of >80% from one day to the other.
Hotspots that normally witness beacons several times per hour. Now only get rewarded several times per day.
What these users have in common:
1 ->
This could be a logical consequence since DIY users can see the miner logs and learn what is happening.
2 ->
The issue is not tied to one brand. Even when reverting to original manufacturer image does not solve the issue. The local network is OK as tested with helium status and other networking tools. The issue is both with relayed and non-relayed hotspots. Mostly DIY setups but there are reports of original manufacturer hotspots as well.
The affected hotspots do see a high amount of witnesses throughout the day. This is confirmed using the miner logs.
It is not a classical case of p2p issue for the hotspot not being able to dial the challenger. The logs clearly show a successful submission of PoC witness to the challenger.
There is currently a lively discussion ongoing on #diy-packet-forwarder and #poc-discussion on the Helium Discord.
Unfortunately no one from the Helium team has chimed in.
Affected hotspots still do send out challenges. However the result is usually a low number(0-6 witnesses) indicating that challengees cannot reach the challenger.
"Add ability to control peer gossip for debugging purposes"
This new piece of code could be related.
When the gossip is disabled, rewards will slowly die down as the miner cannot be found in the network.
Are hotspots selectively grey listed? Like a HIP40 deny list?
helium/erlang-libp2p@853cada
This is how the reward drop looks like:
Miner log, most of these PoC submissions were not rewarded:
The text was updated successfully, but these errors were encountered: