My hard disk is starting to show its old age. On my Ubuntu system I 
occasionally get in a state where the root filesystem is switched to read 
only mode.

Today this happened during a "git fetch" (or maybe a "git rebase") that 
triggered a "git gc".
After reboot+fsck, my git repo is corrupted.

"git gc" fails ("fatal: unable to read <sha1>").
"git fsck" reports missing blobs and dangling commits.
"git status" reports I have deleted files but I haven't (this looks like a 
failed checkout), and a "git checkout" of those files fails because git 
fails to find the blob.

I hope that the missing blobs are on the remote (git+ssh).
How can I tell git to redownload all objects linked from HEAD of the 
remote, bypassing what it thinks it already has?
As "git fsck" is reporting "dangling commits", may be I could dump those 
commit to find out if the missing stuff is important (committed recently by 
myself) or not (probably on the server)?
Maybe there is way to unpack packs? I did another clone from the server, 
but I only get packs in .git/objects. If I could explode them, maybe I 
could repair by corrupted repo...
I don't know how to do any of those things...

If I have to go from scratch from a new clone, I would at least be able to 
import my stash from the corrupted repo...

I'm using git 1.9.1.


You received this message because you are subscribed to the Google Groups "Git 
for human beings" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to git-users+unsubscr...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Reply via email to