From: | Bruce Momjian <bruce(at)momjian(dot)us> |
---|---|
To: | Craig McIlwee <craigm(at)vt(dot)edu> |
Cc: | Simon Windsor <simon(dot)windsor(at)cornfield(dot)me(dot)uk>, pgsql-general(at)lists(dot)postgresql(dot)org |
Subject: | Re: Database upgrade Pg11 to Pg12 issues |
Date: | 2021-02-04 03:29:35 |
Message-ID: | 20210204032935.GA3887@momjian.us |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
On Wed, Feb 3, 2021 at 10:07:03PM -0500, Craig McIlwee wrote:
>
> (replying to the entire list instead of Bruce only this time...)
>
>
>
> This doesn't make sense to me. Since we hard-linked, why would 12 be so
> much smaller? If it was symlinks, I could imaging that, but it doesn't
> use symlinks, just hard links, so it should be similar. Please look at
> the size of main/base on both, since that is where your data is. Maybe
> 11 just has a lot of old WAL that isn't copied.
>
>
>
> It depends on how the directory sizes were measured. "du" won't count the same
> file more than once, so if space was measured using one of the following
> commands:
>
> du -hs /var/lib/postgresql/
> du -hs /var/lib/postgresql/11/main /var/lib/postgresql/12/main
>
> Then you would see a large value for the 11 directory and then the 12 directory
> would only show the files that weren't already counted when measuring the 11
> directory. Running du on each version's directory separately might give
> results that are more in line with expectations.
Yes, I was assuming the du commands were run separately on each
directory.
--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com
The usefulness of a cup is in its emptiness, Bruce Lee
From | Date | Subject | |
---|---|---|---|
Next Message | Niels Jespersen | 2021-02-04 05:03:57 | SV: SV: Npgsql and the Connection Service File |
Previous Message | Craig McIlwee | 2021-02-04 03:07:03 | Re: Database upgrade Pg11 to Pg12 issues |