On Tue, Apr 15, 2014 at 11:58:32AM -0400, Nigel Horne wrote:
> On 15/04/2014 10:40, Peter Pentchev wrote:
> >On Tue, Apr 15, 2014 at 10:19:23AM -0400, Nigel Horne wrote:
> >>OK, so now can you do the following for me?
> >>
> >>1. Still, let's see the output of netstat -atpn | fgrep -e :2525
> >>
> >>tcp        0      0 0.0.0.0:2525            0.0.0.0:* LISTEN
> >>3123/stunnel4
> >>>2. Don't kill the running stunnel process yet, I'd like to check what's
> >>>    going on.
> >>>
> >>>3. Show me the output of:
> >>>    ls /var/run/*.pid
> >[snip no stunnel.pid or stunnel4.pid in /var/run/]
> >>>    ls /var/run/stunnel/*.pid
> >>ls: cannot access /var/run/stunnel/*.pid: No such file or directory
> >>>    ls /var/run/stunnel4/*.pid
> >>>    (it's okay if the second command gives back an error, I expect that;
> >>>    it's also kind of okay if the third one gives an error, but I'd still
> >>>    like to see that :))
> >>ls: cannot access /var/run/stunnel4/*.pid: No such file or directory
> >OK, now this is a bit weird.  I'll have to check again with an older
> >version of stunnel, see how it manages its pid files.
> >
> >What you should do now is stop the running stunnel process by killing it
> >directly:
> >
> >   kill 3123
> >
> >Then "service stunnel4 start" should properly start the new stunnel
> >process.
> 
> It did.
> 
> Thanks.

OK, thank you for the assistance and the confirmation!

Laszlo, could you take a look at the  Yet Another Debian Revision
Update(tm) I've pushed to the collab-maint repo that reenables the
"create a pid file by default" functionality?  This will help people
who:

- use stunnel mainly to manage tunnels that are started automatically
  through /etc/stunnel/*.conf files
- only have a single /etc/stunnel/*.conf file with no pid file setting
  or have multiple /etc/stunnel/*.conf files, but exactly one of them
  has no pid file setting and uses the default /var/run/stunnel4.pid

In 5.01-1 and 5.01-2, the init script would be incapable of monitoring
this configuration - it would start it, but then it would not be able to
tell that it has been started ("status" would say "stopped" and "start"
would attempt to start it again and fail).  Now, in 5.01-3, the init
script will be able to monitor the spawned process (since stunnel will
create a pid file by default), but it will also warn about it.  I think
that in a future version (but possibly after jessie is out) the init
script should refuse to start without a "pid" setting in the *.conf
file, so that we can revert to upstream's decision to not create a pid
file by default.  This, in its turn, will help people who run stunnel in
pretty much *any* other way - start it by hand, run it through
daemontools, etc - since they now have to specify a pid file even if
their startup mechanism does not need one at all.

Time's a bit short, but it might be a good idea to upload this version
in the next couple of days before stunnel4 3:5.01-2 migrates to testing.
Of course, the alternative is to bump this bug's severity so it blocks
the migration.

G'luck,
Peter

-- 
Peter Pentchev  r...@ringlet.net r...@freebsd.org p.penc...@storpool.com
PGP key:        http://people.FreeBSD.org/~roam/roam.key.asc
Key fingerprint 2EE7 A7A5 17FC 124C F115  C354 651E EFB0 2527 DF13
This sentence contains exactly threee erors.

Attachment: signature.asc
Description: Digital signature

Reply via email to