Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 10 Jun 2010 11:28:23 +0200
From:      Harald Schmalzbauer <h.schmalzbauer@omnilan.de>
To:        freebsd-stable@freebsd.org
Subject:   tmpfs problem [Was: Re: extremely strange results with "mail" or "make" after one day uptime]
Message-ID:  <4C10B037.6030702@omnilan.de>
In-Reply-To: <4C10ADAD.2010908@omnilan.de>
References:  <4C109DF7.6070007@omnilan.de> <4C10ADAD.2010908@omnilan.de>

next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enigBD6362A107F1EEB5AE41A1A7
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Content-Transfer-Encoding: quoted-printable

Harald Schmalzbauer schrieb am 10.06.2010 11:17 (localtime):
> Harald Schmalzbauer schrieb am 10.06.2010 10:10 (localtime):
>> Hello all,
>>
>> I have absolutley no idea how it comes:
>> After one or two days of uptime, 'make -j3 buildworld' just returns=20
>> without any error but without doing anything.
>> Also sending mail via 'mail' produces mutilatet output: no valid rcpt,=
=20
>> no subject.
>> When I reboot the machine everything is fine again.
>> Daily status reports also stop working; that's how I found this=20
>> absolutley mysterious spookie thing.
>> And: sshguard destroys hosts.allow. It's suddenly empty.
>=20
> It seems that any text handling routine gets crazy because=20
> /var/run/jail_XXX.id is also empty when starting new jails.
> It worked at machine boot, since older running jails do have a number i=
n=20
> their id file. Only newer started/restarted jails have an empty id file=
=2E
> Also the hosts.allow gets reproducable emptied by sshguard.
>=20
> Where's the part of FreeBSD doing such text manipulation?
> Maybe that's also responsable for makefile parsing and such explains th=
e=20
> 'make' failure? Interesting is that 'make' without -j3 at least starts =

> the build, but reproducable fails at different points while src tree is=
=20
> absolutely consistent. If I mount it elsewhere everything compiles fine=
=2E
>=20
> And to ephisize: This mysteric behaviour of some base system components=
=20
> appears only after some uptime. I haven't seen it the first day after=20
> fresh booting.
>=20
> Anybody any ideas? I'm desperate because I don't know here I could star=
t=20
> to search/test.

Ok, luck seems to be with the stupid this day ;)

I identified tmpfs as the culprit.

'head /etc/hosts.allow' correctly returnes two lines!
'head -n 2 /etc/hosts.allow' > /tmp/test results in a empty file
'head -n 2 /etc/hosts.allow > /var/tmp/test' produves a file with the=20
expected two lines.

Ok, next time I'll better adhere to developers experimental warnings...

Is there anything I can do for the hackers? I haven't had that symptom=20
before on any other machine and I'm using tmpfs for quiet some time.=20
Maby it's an interesting edge case.

Thanks,

-Harry


--------------enigBD6362A107F1EEB5AE41A1A7
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.13 (FreeBSD)

iEYEARECAAYFAkwQsDcACgkQLDqVQ9VXb8gxfwCeL3FHajyKCGRj9tUJNENNlYtK
UWsAn3L1ux1OJ61TEmwV8IqezzMnb848
=AWZs
-----END PGP SIGNATURE-----

--------------enigBD6362A107F1EEB5AE41A1A7--



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4C10B037.6030702>