Log is weird: spam "Looking up qualities took too long" #4803
Replies: 5 comments
-
I have enabled debug logging mode and investigated further. Looks like my usual lookups are 0-3 seconds, like
so it should be fine. But I was able to catch the issue, and it looks like this (log below). Can you please help? Where do I look further to eliminate this? Is this maybe a bug?
|
Beta Was this translation helpful? Give feedback.
-
Signature points usually come one every ~10 seconds. But sometimes they come in bursts of 5-30+ at the same second. Here is an example (log below). Also, this is how it looks in GUI. A lot of "last attempted proof", the same exact second, and as you can see even with different "plots passed filter", looks like this attempted proofs are not duplicates but different? Although I've read Papers, I have no real technical understanding of Chia protocol. Can you please help, is this normal to get 30x different new_signage_point_harvester during the same second? It feels like it's not normal. I'll wait for your feedback for a day more, and elevate this to a bug. Thank you.
|
Beta Was this translation helpful? Give feedback.
-
I'm having the same problem. Looks like the plots re-loading causes the issue. The problem goes away when I increase (running 5000 plots on Mac Mini M1, with FullNode v1.1.7) |
Beta Was this translation helpful? Give feedback.
-
Systems do have a limit, not only turing machines: the more you higher the average load, the less you can react to peaks. Overloading a box with storage leads to average acceptable time on average load, and to unacceptable response time on load peaks. Try to split boxes and/or to configure them to stay as much idle as possible. |
Beta Was this translation helpful? Give feedback.
-
@dima-stefantsov , since this ticket was opened a number of years ago without further comments it seems this issue has been resolved. The best place to reach our support team is on Discord (https://discord.gg/chia) or by reopening this ticket. |
Beta Was this translation helpful? Give feedback.
-
UPD: debug log shows that it starts with simultaneous 20x new_signage_point_harvester at the same time. I've elaborated in first comment below.
Hi. After @codinghorror post, you have added "lookup too long" in log as WARNING. It is important to keep it under control for me.
I'm sometimes getting this warnings, but what I get is weird, I don't know how to interpret it. I only have guesses how your harvester actually works. Here is what I get: (log below)
Most notable thing here is timestamp of log records - it just comes in bursts, notice all this lines being written within 50ms.
So how does your harvester and logs work, what exactly is going on?
Notice fast local drives have "less than 1ms max", yet they still managed to get onto the slowlog with up to 55 seconds lookup. Something weird is going on, it should never happen!
I'm on Windows 10
Using GUI app 1.1.5
Beta Was this translation helpful? Give feedback.
All reactions