linux-ext4 - Re: journal_async_commit and external journal

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <7DBBC106-3E65-4AA2-A460-E48D05FE220F@dilger.ca>
Date:   Mon, 7 May 2018 22:54:36 -0600
From:   Andreas Dilger <adilger@...ger.ca>
To:     Niklas Hambüchen <mail@....me>
Cc:     Ext4 Developers List <linux-ext4@...r.kernel.org>
Subject: Re: journal_async_commit and external journal

On May 4, 2018, at 9:04 PM, Niklas Hambüchen <mail@....me> wrote:
> 
> Hello,
> 
> The article https://raid6.com.au/posts/fs_ext4_external_journal_caveats/
> is an often-cited source about setting up external journals.
> 
> It claims that without the `journal_async_commit` mount option ext4 with
> external journal will eventually go read-only.
> 
>> Apparently external journal can't be updated in synchronous manner so
>> sooner or later ext4 switches to read-only mode following journal commit
>> error.
> 
> https://www.arm-blog.com/ext4-journal-nvram-card/ makes the same claim.
> 
>> The “journal_async_commit” is mandatory in order to use the external
>> journal. If you forget to mount the filesystem with this flag it
>> eventually become read-only and will need a diskcheck before you can
>> mount it again! This problem occurs because the external journal cannot
>> sync in synchronised mode.

This second report is just a repeat of the first one (it even links to the
earlier article).

> But the only upstream bugs I could find are records of situations where
> _with_ the option ext4 went read-only at boot or remount.
> 
> Is the issue described in these articles a known problem?

In any case, I couldn't find any reports of this causing problems on the
linux-ext4 mailing list, which is where it should be reported if anyone
has a problem with ext4 (or bugzilla.kernel.org if necessary).

If you have the time and a system to test it, you could run with this config
with journal_async_commit disabled for some time (with plenty of reboots)
and see if you find any problems.  That said, if enabling journal_async_commit
works for you, then that is the preferred option anyway, since it adds
checksums to the journal blocks (which makes for safer recovery), and is also
faster.

Cheers, Andreas

Download attachment "signature.asc" of type "application/pgp-signature" (874 bytes)