search

h9-dev / chia-miner

The miner for Chia.
https://www.h9.com
225 stars 40 forks source link

Log bad plot filename to console #618

Open ghost opened 3 years ago

ghost commented 3 years ago

How to find which plot is bad?

chia plots check says all is OK

but not hpool-miner:

goodbit, failbit, badbit, eofbit: 0010
exception : badbit or failbit after reading size 2 at position 108492278296
WARN[2021-06-11T21:22:52Z] prove invalid                                 file=loggers.go func=logging.CPrint line=162 tid=12
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107537296643
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107558405292
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107507334100
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107516665506
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107538870186
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107529965283
goodbit, failbit, badbit, eofbit: 0010
GetQualitiesForChallenge exception : badbit or failbit after reading size 4 at position 107521970965
mmaa-crypto commented 3 years ago

have you opened gui and console at same time ??? close both and open console singly ; its gonna be ok i think

ghost commented 3 years ago

have you opened gui and console at same time ??? close both and open console singly ; its gonna be ok i think

No, I do not have GUI at all, it is linux console.

It is only 1 plot bad. Always the same with the same numbers. But not clear which one.

It is faster to replot it than clarify why official plots check says all is OK.

ghost commented 3 years ago

It is most likely bug in software. I expect race condition. It looks like miner wants to open file which is already locked by itself.

It happens on rescan of directories.

room101-dev commented 3 years ago

It is most likely bug in software. I expect race condition. It looks like miner wants to open file which is already locked by itself.

It happens on rescan of directories.

[ Rescan is useful in plotting machines, where you have new plots being created on local machines every hour, ... otherwise if your plots are static then it should be off. If you plot and don't have it enabled, then you need to start/stop the client when new plots are created. ]

I think this is why they have 'rescan' turned off by default in the config.yaml, while I like the 'rescan' if your plotting, if your plots are fixed, then better to keep this off.

WRT the 'failbit' problem, my experience is that you have a (dot) .plot file that is not full-length, go to the linux gui folder-browser, sort all plots by size, and find one that is less than 108gb, and remove it, then you will be good. Why this mal-formed 'short' plots work with 'chia plots test', well like all chia-net sw it doesn't work, while HPOOL has real problem with 'fake plots' they actually check for things like 'complete length' ( real correct w-32 plot )

prinze77 commented 3 years ago

You can get this error if your disk got disconected Try to unmount and mount again the disk

ghost commented 3 years ago

You can get this error if your disk got disconected Try to unmount and mount again the disk

It is not USB disk. And there no messages in dmesg about any issue.

prinze77 commented 3 years ago

When I got this issue my disk was a sata hdd disk, I powered the disk off then powered it on mounted it back and the error disappeared

ghost commented 3 years ago

When I got this issue my disk was a sata hdd disk, I powered the disk off then powered it on mounted it back and the error disappeared

It is hardware RAID array, each issue with disk is logged by controller. There are no errors.