Signage Point 'Bursts'?

Hey Everyone,

Looking for feedback on my current setup that seems to be working correctly, but there are these odd groupings of signage point communications to and from my full node to my remote harvester. Typically they come in pairs where the signage point is written out (local machine and remote harvester) and then receives likewise paired responses.

Sometimes they come in large bursts of 5+. Is anyone else seeing this issue? I’ve found some threads about it but with no workaround or clear answer as to why or if it even matters. Including some logs below.

192.168.1.16 is my remote harvester.

EDIT: [BUG] multiple new_signage_point spam, looks like malicious network attack · Issue #5410 · Chia-Network/chia-blockchain · GitHub Adding an associated issue on the repo

2021-05-20T12:35:30.953 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 127.0.0.1 372910623295859b822a3d5c40046b2246b924717973fdecebf3823d2ff194da
2021-05-20T12:35:30.954 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 192.168.1.16 ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209
2021-05-20T12:35:30.956 farmer farmer_server              : INFO     <- new_signage_point from peer 49b5948bd874c61cdc0281389ea298d131aa70f363adce6ef68ed8be9436e269 127.0.0.1
2021-05-20T12:35:30.957 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 127.0.0.1 372910623295859b822a3d5c40046b2246b924717973fdecebf3823d2ff194da
2021-05-20T12:35:30.959 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 192.168.1.16 ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209
2021-05-20T12:35:30.960 farmer farmer_server              : INFO     <- new_signage_point from peer 49b5948bd874c61cdc0281389ea298d131aa70f363adce6ef68ed8be9436e269 127.0.0.1
2021-05-20T12:35:30.961 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 127.0.0.1 372910623295859b822a3d5c40046b2246b924717973fdecebf3823d2ff194da
2021-05-20T12:35:30.962 farmer farmer_server              : INFO     -> new_signage_point_harvester to peer 192.168.1.16 ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209
2021-05-20T12:35:31.555 farmer farmer_server              : INFO     <- farming_info from peer ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209 192.168.1.16
2021-05-20T12:35:31.632 farmer farmer_server              : INFO     <- farming_info from peer ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209 192.168.1.16
2021-05-20T12:35:31.858 farmer farmer_server              : INFO     <- farming_info from peer ad87555074ab9147af1a7c193bffb3f9233b56c479f5c50721844a576593e209 192.168.1.16
2 Likes

To add to this, I think I may have found a correlation. Whenever this seems to happen, one of my start_full_node.exe is hitting my CPU suddenly hard. It spikes from what is usually in the 2% up to almost 30+. This is on an i9-9900k. I’ve just tried to set the process to a higher priority and also cutting back on other processes that are eating up CPU resources to see if it helps at all.

Otherwise, something weird is going on with information being passed p2p would be my guess?

1 Like

I’ve been trying to debug the same issue. I’ve been running chiadog and the bursts of signage points seem to correspond with the full node reporting skipped signage points, followed by a warning the harvester did not participate in any challenges for n seconds (typically around 60-100 seconds).

I actually ended up switching my full node to a different/more powerful machine rather than having a full node/everything on a less powerful xeon server, but I’m able to replicate this issue in the different setup also (both linux, different distros).

This issue is the same I think, and it seems that some people are having this issue regardless of how powerful the system is.

I only started running chiadog and chiaharvestgraph for around 24 hours also (which indicates underfarming) so I’m wondering if this is what has caused my actual rewards to be between 2-4x the expected time over the last month (ETW of 4-7 days and no rewards in 20 days with a system that appears working in all ways- perhaps coinciding with updating to 1.1.5?) and just like the responding to challenges in 30 seconds issue there is actually something seriously wrong and it’s just not being surfaced anywhere in logs.

2 Likes

From the 1.1.6 release notes:

Fixed issues with signage points: duplicate signage point spikes, and bugs in signage point cache.

Will see if this makes any difference

I had this happening on my full node / farmer where I was also plotting… After I reduced the number of parallel plots, it went away. Related?

It’s interesting because I was doing the same thing and while I do think/see that if you are asking too much of your full_node machine (aka also plotting a lot), it’s not required to cause it. I recently had to switch to a remote harvester because my NAS search times were taking too long. I figured it had something to do with that originally but have tried every config possible and still see the issue.

I think it’s just not as obvious to people who aren’t running remote harvesters…and honestly that along with not having won a reward in a while made me dig a bit deeper.

Good to see the update. Will try it out now.