From: Ramkumar Ramachandra <artagnon@gmail.com>
Date: Sat, 03 Aug 2013 07:52:25 GMT
Subject: Re: [PATCH v2] gc: reject if another gc is running, unless --force is given
Message-ID: <CALkWK0k1vf_WE=pV5XTMAM6Ax6rL3sXu4qx_eyYwvWGsZjXgmA@mail.gmail.com>
In-Reply-To: <1375510890-4728-1-git-send-email-pclouds@gmail.com>

Nguyễn Thái Ngọc Duy wrote:
> This may happen when `git gc --auto` is run automatically, then the
> user, to avoid wait time, switches to a new terminal, keeps working
> and `git gc --auto` is started again because the first gc instance has
> not clean up the repository.
>
> This patch tries to avoid multiple gc running, especially in --auto
> mode. In the worst case, gc may be delayed 12 hours if a daemon reuses
> the pid stored in gc-%s.pid.

Definitely looks like a good solution. Thanks for this.

I'm currently on vacation, so can't apply and test: sorry.

> +       if (!force &&
> +           (fp = fopen(git_path("gc-%s.pid", utsname.nodename), "r")) != NULL &&
> +           !fstat(fileno(fp), &st) &&

It's open for a very short period of time, so lockfile (which we'd
normally use) would probably be an overkill.

> +           time(NULL) - st.st_mtime <= 12 * 3600) {

Quick question: is this kind of file-lifetime used anywhere else in git.git?

> +                       if (auto_gc)
> +                               return 0; /* be quiet on --auto */
> +                       die(_("gc is already running"));

Nice.