Write-hashtbl hashstore: collision detected

Node Operators troubleshooting
,
1

Hi, I’m using write-hashtbl and run into a new situation:

hashstore: put:{key:1b1bbd32f020502e52e839717e13a486ae0b9f6eff7c2f25001bf1df606bc12e offset:1073162112 log:116 length:2319872 created:20450 (2025-12-28) expires:0 (1970-01-01) trash:false} != exist:{key:1b1bbd32f020502e52e839717e13a486ae0b9f6eff7c2f25001bf1df606bc12e offset:95672576 log:100 length:2319872 created:20450 (2025-12-28) expires:0 (1970-01-01) trash:false}: hashstore: collision detected
	storj.io/storj/storagenode/hashstore.(*HashTbl).insertLocked:384
	storj.io/storj/storagenode/hashstore.(*HashTblConstructor).Append:525
	main.(*cmdRoot).Execute.func2:110
	main.(*cmdRoot).iterateRecords:201
	main.(*cmdRoot).Execute:109
	github.com/zeebo/clingy.(*Environment).dispatchDesc:129
	github.com/zeebo/clingy.Environment.Run:41
	main.main:29
	runtime.main:283

How do I fix this?

2

Seem like I just need to delete log 64 (decimal 100) and log 74 (decimal 116) it will work again.

3

This is not a solution. It’s curing headache with a guillotine. The solution would be to find a root cause and fix that, not kill the patient.

2 Likes
4

This is not a dangerous error. Old piecestore silently ignored double uploads, hashstore doesn’t.

Should be very rare (for example: piece is deleted, and repair tries to re-upload with the same ID).

never delete log files because this. That will delete healthy pieces and will cause disqualification.

1 Like
5

It look like a dangerous error, effectively stop node from being online.. how do we properly fix that?

P/s: since I run into this issue first, may I provide node id to prevent my node from being disqualified here?