The underlying problem has been fixed in rc.subr and no longer requires
rc script customization to get a sane behavior by default.
http://cvsweb.openbsd.org/cgi-bin/cvsweb/src/etc/rc.d/rc.subr?rev=1.99&content-type=text/x-cvsweb-markup
signature.asc
Description: OpenPGP digital signature
__
On Tue, Jun 23, 2015 at 01:09:07PM +0200, Fabian Keil wrote:
> In the ElectroBSD version of the port I recently went with the
> mentioned "specify all log files in the torrc" strategy ...
You might also enjoy the --defaults-torrc option, which you can
use for giving Tor new defaults while still le
> Date: Tue, 23 Jun 2015 13:09:07 +0200
> From: Fabian Keil
>
> nusenu wrote:
>
>>> …
>>> Did you test with custom flags like "--+Log ${tor_loglevel} file
>>> ${tor_logfile}"?
>>
>> This should read "--Log" no?
>
> While I prefer using neither "--Log" nor "--+Log", the latter
> is valid synta
nusenu wrote:
> >> One line "patch" for /etc/rc.d/tor to address this issue: (tested
> >> with and without custom daemon_flags)
> >>
> >> 8a9,10
> >>> pexp="${daemon}${daemon_flags:+ ${daemon_flags}}$"
> >>>
> >
> > Did you test with custom flags like "--+Log ${tor_loglevel} file
> > ${tor_log
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512
> Did this reply make the intentions clearer?
>
>> Yes, thanks. Though my first inclination would be to tell users
>> "don't use rc.d if you've got the ansible role configured", I
>> suppose it can't hurt to make it foolproof. I'll test your patch
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512
>> what do you think of changing OpenBSD's tor rc script to require
>> perfect process matches when sending signals to them instead of
>> the current "kill everything that starts with..." approach?
>>
>> I've put tor-relays on CC so people can spea
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512
Hi Pascal,
thanks for the fast reply.
I think there is a slight misunderstanding.
>> rc.subr(8) does not have multi-instance support, and it will most
>> likely stay that way.
Correct, and I'm _not_ trying to add that in any way to rc.subr or to
nusenu wrote:
> what do you think of changing OpenBSD's tor rc script to require
> perfect process matches when sending signals to them instead of the
> current "kill everything that starts with..." approach?
>
> I've put tor-relays on CC so people can speak up if this change is not
> desired or
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA512
Hi Pascal,
what do you think of changing OpenBSD's tor rc script to require
perfect process matches when sending signals to them instead of the
current "kill everything that starts with..." approach?
I've put tor-relays on CC so people can speak up