Anyone any Idea what this is???
And is it a problem?
Just updated to 1.29 yeterday…cleared debug log… and now get lots of these in the new debug log…
You should post the full ERROR message (which this is not) for peeps to help.
This appears to be part of a WARNING:
full_node chia.full_node.mempool_manager: WARNING add_spendbundle [long string of numbers] took 1.11 seconds. Cost: 1579784310 (14.362% of max block cost)
Most of the time you do not need to worry about WARNINGs as they are non critical. It is errors that are of concern.
This WARNING is generated when add_spenbundle takes over 1 second. Even as a warning it is over aggressive. I would not worry at all unless it was over 5 seconds.
It is probably an indication that you have some slow response times. Do you have a dashboard like farmr installed to help you monitor your farmer/harvester?
I don’t suspect a big issue but you should probably look at your response times in general.
My response (farmr) stats >0.14772s min, 0.58s ave, 0.40s median. I get this warning continually and constantly in my logs since the recent upgrade. I don’t believe (although I don’t know, or have found anyone who does know) what it means in the first place. Anything to do with response times seems unlikely. Chia docs are of no help. And I’ve asked in several places.
· 6d
This is a new logging mechanisms they implemented. You can compare ~/chia-blockchain/chia/full_node/mempool_manager.py files for old and new version.
In 1.2.8 they only report such messages when time > 1 second but instead of INFO they rate it as WARNING.
One could argue that >1 sec is a bit aggressive. Many people get them, not so much to worry about as long as it’s under, say, 7 sec.
Credits to user neurosis from the SpaceFarmer discord.
1 Like
I’ve seen this asked in keybase support. Not by me.
Was answered today.
Here’s the response.
"if you arent running other stuff, there really isnt much to do, other than upgrade the hardware - but looking at that part of the log, it doesnt look like it is something really pressing, yet. every few hours you got one taking over a second… it should be doable "
The person then asked if it would affect rewards.
Answer below.
“i doubt it does in your case, as long as it is just every few hours, and it is just over 1 second, but keep an eye on it in the future”
…doesnt look like…
…should be doable…
…i doubt…
Those look like guesses, not factual answers. I want someone to explain what the message means in terms understandable to the average farmer, functionally. Not likely, not think, but know. Then I’ll understand clearly it’s importance.
The message: “WARNING add_spendbundle [long string of numbers] took 1.11 seconds. Cost: 1579784310 (14.362% of max block cost)”
1 Like
Well that was info from the official support.
Not sure what else I can say…
Alot of what I’ve seen answered are just issues we’ve covered here in the community anyway.
I maybe shouldn’t break down code into understandable fragments, but I’ve been in crypto a while and know the gist of how some things work, so I try.
Add spend bundle to me is someone trying to add something to be included in the chain.
It’s to do with mempool , which is normally a memory pool of transactions waiting to be added.
I’d guess the numbers are to identify the different transactions.
I’d guess cost is cost of transaction, and its used 14.362% of available block.
If you want more than that, I can only suggest researching online or asking in support exactly what’s going on there.
I can only give a best guess.
They also didn’t clearly answer my query, but said what they thought was happening, not a definite, but I was great full, better than no support.
Ask & u shall receive > from ply, support >
“each node has a “mempool” of open transactions, waiting to be committed to a block. If the farmer has a good proof, it will in cooperation with the node create a block, and include as many transactions as there is either room for, or it runs out of transaction to add from the mempool… when users creates these transactions, they get propagated onto the network, and the nodes keep them in their mempool, for adding to a possible block they create… before adding anything to the mempool, it verifies the transaction is not previously spend, and that it actually controls the coin ID’s it is trying to transfer… it is this verification that took over a second…”
1 Like
The initial answers I showed were also from pyl.
They try to help, but don’t know every issue.
Perhaps, each node’s fault then? So, maybe, it’s a memory and/or CPU thruput issue in our nodes running chia sw? The mempool is in each node… or PC. That’s what I read.
Just looking for someway way to improve it, if possible.
As per the response I posted earlier.
If your machine is not plotting or doing anything else to strain it, maybe upgrade hardware.
I’m really not sure how ppl are farming on pi’s.
Something doesn’t fit. My Ryzen 3600 sits here doing almost nothing as I watch each threads activity, on a gig fiber internet connection on a 2.5gb ethernet network with 3 nodes.
It seems more likely latency somewhere outside my ecosystem is at issue here. Again, something in this WARNING story doesn’t add up.
I’m farming on an machine with an 8350, so 4 cores 8 threads, my cpu is at about 6% normally.
I’ve not gone to latest version yet, but I’ve seen I get the same issue, just there’s no warning triggered yet ( till I update ).
In your response it said each node has a mempool, so it seems that’s def local to you.
Where your node checks it hasn’t already been spent is in question.
It could be your node checking the last x no of blocks to make sure it’s not included ( to me that’s a common sense approach).
But maybe it’s asking for confirmation elsewhere, I really have no clue.
This is not a new message. You are just seeing it now because it was just changed from INFO to WARNING in the last upgrade. IMOP it should not have been. IMOP it should only be a WARNING if it is over 5 seconds and should only be of concern to the user if it is over 5 seconds repeatedly.
Spend Bundle - A spend bundle is a collection of spends grouped together with an aggregated signature to be sent to the network.
I very much think they will relax the timings for that warning once enough ppl flood support questioning how to fix it.
It would be most helpful if a panel of the GUI was given over to error indicators (or user defined error, warning, debug, or all maybe), and a clue as to what they mean to the user, if anything (and ideally, how to possibly rectify.
ROFL!!!
You will love this non-answer thread I found while trying to answer a new question:
Chia devs do not respond and then close thread for lack of response, lolol!
I think a lot of questions like this “Go Stale” as the forum ends it, simply because no one steps up with any answers. IOW, peps quickly learn, “Why ask questions, when no one answers?” And so stop posting.
I’ve seen alot of issues on github that just get closed like that for inactivity.
Best place to ask is keybase imo.
It would, hopefully in time it will happen.