Re: [PATCH v2 3/5] refs/reftable: add fsck check for number of tables

From: Karthik Nayak <karthik.188@gmail.com>
Date: Sep 15, 2025, 09:27 UTC
Message-ID: <CAOLa=ZR9nXkdXuMik2uq8+kpxQwmB72r7XZB0rnJSvKU7mKmkw@mail.gmail.com>
In-Reply-To: <aLf3NyodTZ38N8vY@pks.im>

Patrick Steinhardt <ps@pks.im> writes:

Show 14 quoted lines

> On Tue, Sep 02, 2025 at 09:05:23AM +0200, Karthik Nayak wrote:
>> Introduce a reftable fsck check to check that the number of files in the
>> reftable directory matches the number of files listed in 'tables.list'.
>> We do this by iterating over the files in the reftable directory and
>> counting all the files present excluding the 'tables.list'. This is also
>> exposed over Git's fsck checks as a 'badReftableStackCount' error.
>
> This feels overly strict, as it can always be the case that a concurrent
> process is currently updating the stack. Furthermore, it's expected that
> on Windows systems deletion of an old table may not work because the
> file is still kept open by another process. The reftable library is
> prepared to handle this alright and will re-try deleting the table at a
> later point in time.
>

Yeah that makes sense.

> So maybe a better check would be to verify that there are no files with
> unexpected names in the directory?
>

I was hoping to add structured consistency check in a layered format

- Stack
  - For each `Table`
    - For each `Block`
      - For each `Ref`

But this wouldn't belong to that, since it isn't part of the stack. So, I'll keep the above structure and also add this to the stack level. So we'll have

- Stack
  - For each `Table`
    - For each `Block`
      - For each `Ref`
  - Other Stack level checks
    - Check other files in the repo

> Patrick