Hey there,
strange things started happening after installing V1.2.6:
“Have multiple copies of the plot …”
which is a total b/s since there are zero duplicates of my blocks. My farmer never had this behavior before from April 2021 until I installed V1.2.6.
And no, its not just the GUI, its in the logs too, see screenshot:
Its quite annoying since the plots “degrade” to 50% over 24h, so I have to restart chia at least every 4 hours to have all my 2800 plots eligible for farming. All plots on randomly stated harddisks are all shown as multiple copies of the plot.
Checked and verified several times, there are no dupes at all. BUG seems to hit other users too, see my post @https://chiaforum.com/t/farmer-suddenly-shows-have-multiple-copies-of-the-plot-on-chia-v1-2-6-and-1-2-7/13129/11
OS is Windows Server 2019, Chia V1.2.7, Plots are hosted on several Synology NAS via 10Gbit network.
Cheers
fair enough, you could contact @pyl on our keybase https://keybase.io/team/chia_network.public
or email it to rc 'at' chia.net ?
Done - thanks for looking into that! @loppefaaret
oh wow - that just is completely random, and so odd... could you do me a favor, if you can catch it in this "Multiple copies" state during today, run a chia plots show
command, and triple check that is does not output the same plots dir more than once ?
on windows, that would be under cd ~\AppData\Local\chia-blockchain\app-1.2.7\resources\app.asar.unpacked\daemon\
folder, where you can run .\chia.exe plots show
oh wow - that just is completely random, and so odd... could you do me a favor, if you can catch it in this "Multiple copies" state during today, run a
chia plots show
command, and triple check that is does not output the same plots dir more than once ?
Yes - thats the thing. There is no pattern. Sure...I forward you the according logs when they come up.
@derscheff if we could get you into DEBUG logging, and get you to send additional logs to me, that could help us narrowing down this bug...
either edit config.yaml
and set
log_level: DEBUG
under the farmer:
section
or through the CLI with .\chia.exe configure --log-level DEBUG
and then restart Chia services (.\chia.exe start -r farmer
)
on windows, you can open a PowerShell window, and do:
Get-Content -Wait ~\.chia\mainnet\log\debug.log | Select-String -Pattern 'multiple'
this will start scanning the debug.log
file for the word "multiple" - as soon as it happens, it will be listed in this window, and you can send the log to me again
Thank you
Hi there, I'm working on loppefaaret on this. Right now, I think this log entry is spurious, in that it is being printed in error, but should not negatively affect any farming. What I do see is some flaky connection to your NAS, which may be more responsible for your issues. For example:
Very long lookup times (almost 30 seconds in some cases)
Plots being removed (I assume you aren't deleting any - but in this case during the refresh, the harvester code could not detect the file (python pathlib exists() failed)
refresh_batch: loaded_plots 0, loaded_size 0.00 TiB, removed_plots 219, processed_plots 0, remaining_plots 0, duration: 3.68 seconds
Those plots then come back later on
refresh_batch: loaded_plots 158, loaded_size 15.64 TiB, removed_plots 0, processed_plots 158, remaining_plots 0, duration: 12.78 seconds
We continue to look into this
I have reproduced one scenario where this could happen - and it does result in those plots getting ignored until the harvester is restarted.
The problem I can reproduce deals with file modification times - the harvester checks the file modification time to see if the file has been changed since it was last seen. I believe in some cases (like maybe NAS or USB) the modification time doesn't match anymore and this causes the plot to get marked as a duplicate and not loaded.
I can simulate by using touch -m
to manually change the modification time of the plot file
@derscheff if we could get you into DEBUG logging, and get you to send additional logs to me, that could help us narrowing down this bug...
either editconfig.yaml
and setlog_level: DEBUG
under the
farmer:
section
or through the CLI with.\chia.exe configure --log-level DEBUG
and then restart Chia services (.\chia.exe start -r farmer
)on windows, you can open a PowerShell window, and do:
Get-Content -Wait ~\.chia\mainnet\log\debug.log | Select-String -Pattern 'multiple'
this will start scanning the
debug.log
file for the word "multiple" - as soon as it happens, it will be listed in this window, and you can send the log to me again
Thank you
Hey - I just read your answer. I can surely do this over the weekend. Will mail it to rc afterwards.
Cheers
@derscheff Can you try the build here: https://github.com/Chia-Network/chia-blockchain/actions/runs/1271511224
With that build I'm hoping the following:
Have multiple copies
warningsIgnoring modification time mismatch
@derscheff if we could get you into DEBUG logging, and get you to send additional logs to me, that could help us narrowing down this bug...
either editconfig.yaml
and setlog_level: DEBUG
under the
farmer:
section
or through the CLI with.\chia.exe configure --log-level DEBUG
and then restart Chia services (.\chia.exe start -r farmer
)on windows, you can open a PowerShell window, and do:
Get-Content -Wait ~\.chia\mainnet\log\debug.log | Select-String -Pattern 'multiple'
this will start scanning the
debug.log
file for the word "multiple" - as soon as it happens, it will be listed in this window, and you can send the log to me again
Thank you
Doing this now. Gonna send the "window report" out tomorrow...
Hi there, I'm working on loppefaaret on this. Right now, I think this log entry is spurious, in that it is being printed in error, but should not negatively affect any farming. What I do see is some flaky connection to your NAS, which may be more responsible for your issues. For example:
Very long lookup times (almost 30 seconds in some cases)
Plots being removed (I assume you aren't deleting any - but in this case during the refresh, the harvester code could not detect the file (python pathlib exists() failed)
refresh_batch: loaded_plots 0, loaded_size 0.00 TiB, removed_plots 219, processed_plots 0, remaining_plots 0, duration: 3.68 seconds
Those plots then come back later on
refresh_batch: loaded_plots 158, loaded_size 15.64 TiB, removed_plots 0, processed_plots 158, remaining_plots 0, duration: 12.78 seconds
We continue to look into this
Could be a hint, but never was a problem in months of farming though
@derscheff Can you try the build here: https://github.com/Chia-Network/chia-blockchain/actions/runs/1271511224
With that build I'm hoping the following:
- No
Have multiple copies
warnings- replaced with
Ignoring modification time mismatch
- Plots not being "lost"
Happy to do so after the 24h peroid of window logging of @loppefaaret ;)
Owner Name | Chia-Network |
Repo Name | chia-blockchain |
Full Name | Chia-Network/chia-blockchain |
Language | Python |
Created Date | 2019-07-16 |
Updated Date | 2022-08-13 |
Star Count | 10914 |
Watcher Count | 340 |
Fork Count | 1913 |
Issue Count | 343 |
Issue Title | Created Date | Updated Date |
---|