Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 29 Mar 2016 04:01:07 +0200
From:      Bernd Walter <ticso@cicely7.cicely.de>
To:        =?iso-8859-1?Q?Jos=E9_P=E9rez?= <fbl@aoek.com>
Cc:        Nikolai Lifanov <lifanov@mail.lifanov.com>, freebsd-arm@freebsd.org
Subject:   Re: Official images without noatime
Message-ID:  <20160329020107.GC68225@cicely7.cicely.de>
In-Reply-To: <6c15a205f6d5126c7d468bd2605be769@mail.yourbox.net>
References:  <mailman.41.1458993601.86944.freebsd-arm@freebsd.org> <4b23b28ffae59216b5dde8f28f665330@mail.lifanov.com> <813ba9c4a1474478daa86fe685acec21@mail.yourbox.net> <56F96C46.80705@mail.lifanov.com> <6c15a205f6d5126c7d468bd2605be769@mail.yourbox.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Mar 28, 2016 at 08:43:55PM +0200, José Pérez wrote:
> Hello Nokolai,
> 
> El 2016-03-28 19:39, Nikolai Lifanov escribió:
> >A simple case is during install /usr/bin/cmp is ran to compare two
> >files, atime for /usr/bin/cmp is updated during a crash, and
> >/usr/bin/cmp is gone on next boot. I then have to copy it out of
> >/usr/obj and into place and run installworld again. It's the handful of
> >utilities actually *used* by installworld that do this and mounting 
> >root
> >with noatime stops this from happening.
> 
> I suspect you have a problem somewhere else, maybe a faulty flash?

Well - unless you don't have a power loss having atime on just
increases the write load.
And having a power loss during install is always a bad thing to happen,
since the card is writing anyway.
Of course chance that you loose the used binary is obvious, since it is
updating the inode of that binary, but the way flash cards work you may
loose way more than that.

> Can you reproduce with another hardware? Can you help us reproduce it?
> What do you mean "atime is updated during a crash"??

If it is crashing (instead of having a power loss) during writes the
situation is different.
Since the card had no power loss all data that the OS had written should
be on the card.
That means either the old inode or the new inode.
First thing to see is why it crashes.
Another good thing to do is to do a full fsck of the filesystem.
Trusting a journal can be a bad idea, since the whole media technology
can't be trusted to begin with.

> >>My RPI2 shutdown cleanly with shutdown(8) or reboot(8).
> >>
> >
> >It doesn't stay down if the power cable is still connected.
> 
> RP does not power down, you have to disconnect the power cable.
> But first you have to stop the OS:
> shutdown -p now
> does that for you (which includes syncing and unmounting disks).

Chances are that your system crashes on shutdown, so that it
automatically reboots.
The only way to find out what happens is plugging something to the
console and see.
That said: it shouldn't crash, neither on shutdown nor when installing
something.
You should hook up a console to find out anyway.

-- 
B.Walter <bernd@bwct.de> http://www.bwct.de
Modbus/TCP Ethernet I/O Baugruppen, ARM basierte FreeBSD Rechner uvm.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20160329020107.GC68225>