Heavy traffic observed

because streaming was exceptional troublesome today i did slow down my node monitoring

but i think it wasn’t the monitoring - i think just found out today that doing a “safe wallet receive” does put some (for my internet) serious load onto the wires for about 40 seconds (red btw is the upload … not sure why that would interfere with my streaming that heavily … but the client for sure is not light on network resource usage)

Was looping that for the last day to try and receive a coin transfer that seems to be exceptional hard to receive… @qi_ma I did send you a message via discord… Not sure this is something you want to investigate or consider a non problem in the future anyway… (the failing receive… Not the network load on receiving)

5 Likes

thx for the info.

would like to have an investigation on it.
could you pls send me the client log you got during the heavy traffic usage transfer?
thx.

3 Likes

is it the same client log you sent me via discord?

1 Like

no - the other log was the send - will send you the attempt to receive too now

3 Likes

Hi, @riddim

I checked the two logs you shared, here here is my summary of what happened:

  • according to the sending log, there was 544400 nanos collected from 91 cash_notes to be transferred to one receiver.
  • This results in a complexed transfer to be composed, which has 91 inputs.
  • And for each of that input, a spend record needs to be published to the network to claim that amount of nano is spent. And each of that record has to hold that complexed transfer.
  • This result in the total traffic involved in send out & receive that tranfer will be aroud couple of hurnderds MB.
  • the send log does show all the 91 spends got sent out and stored by the network. which means your nanos are there in the network.
  • however, the receive log shows during the verification of those 91 spends, there is spend cann’t be fetched, which failed the trasnfer. the receive log does show less peers discovered compared with the send log (131 vs. 220), which might be not with a good connection or in bad time?
  • All these 91 spends have to be ALL fetched to make the trasnfter to be received successfully, which is really a challenge.
4 Likes

thank you very much for taking your time to look into it @qi_ma !

hmmm - i receive from my home machine while i send from a could instance - this is pretty plausible …

… now i feel reminded on

“Christ, that’s not an octopus, that’s a
Cthulhu merge”

…but it’s the only way node runners will be able to “cash in” on their earnings in the future … and bundling 91 cash nodes doesn’t look like an insane amount to me tbh … just thinking loud … (ofc a workaround would be to do a send-to-oneselve on collecting the nanos every 20 nodes or so … which would result in no more than 20 inputs to ever occur in the process …)

2 Likes

yeah, agreed.
there is work to resolve this, and hopely with the next release, this issue shall be resolved

6 Likes

but a reissue once in a while on collection might be a good idea … not having too many inputs seems reasonable somehow

2 Likes

Are we getting, at some stage in future, the ability to specify the wallet a node is to use to receive payments? This would seem to solve having to run a script to send from the individual node wallet to a central wallet.

Honestly this is needed if we want people to run nodes and not have a headache to use their earnings

6 Likes

This exists already in the form of reward forwarding for beta, shouldn’t be a big leap?

4 Likes

Dunno. Hopefully, but I’d like it acknowledged that they will be doing this at some stage

4 Likes

but then i would specify my central wallet as receiver for all my nodes …would have incoming rewards from not only 91 locations but from a couple thousand nodes i’m currently running into my central wallet … and when i try to spend them i’d bundle the cash-notes there and see the same problem … just on a way way larger scale …

maybe a receiving wallet should do a reissue automatically after having received “too many” cash-notes to prevent this?

1 Like

yeah, there is discussions on that,
and change reward_forwarding key does can support that and being an option.

However, need to cross check with other colleagues, as it needs support from node_manager/launch_pad as well.

9 Likes

I have also observed traffic spikes together with heavy cpu spikes, don’t know since last upgrade but have monitored from before that.

Nodes would stay around 50% and then spike up to 100% for seconds almost fully freezing and locking the PC.