Re: fatal: unable to read <hash>
- In reply to: Bjoern A. Zeeb: "Re: fatal: unable to read <hash>"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Wed, 09 Aug 2023 13:03:04 UTC
On 2023-08-08 20:31, Bjoern A. Zeeb wrote: > On Mon, 7 Aug 2023, Marc Branchaud wrote: > >> >> On 2023-08-07 01:18, Bjoern A. Zeeb wrote: >>> >>> % git fsck --name-objects --connectivity-only --no-dangling >>> error: b9cdc058e6eb1b3d8b5e29ad9b911d3da98f65a7: invalid sha1 pointer >>> in resolve-undo of .git/worktrees/bz_maxmodname_D32383/index >>> Checking connectivity: 4743571, done. >>> Verifying commits in commit graph: 100% (384308/384308), done. >>> >>> The good news upfront; I could ditch that worktree and branch if that >>> will fix things. It has follow-up work to the 2021 things but I have >>> a copy of that outside that git tree (I love patch files in port >>> trees :) >> >> If you no longer need the bz_maxmodname_D32383 worktree, then it >> sounds like the easiest solution would be deleting it -- by running >> git worktree remove bz_maxmodname_D32383 >> in your "main" repo (which you created with "git clone"). >> >> (The bz_maxmodname_D32383 branch seems to be OK, according to fsck.) > > And how would one gotten rid of the invalid sha1 pointer in that case? It depends on the corruption. The SHA1 IDs are used at all levels in Git -- to identify commits, "trees" and "blobs" (files). > I removed the worktree and gc now finishes. Glad you worked it out! M.