daemondo fails to launch unbound daemon at boot, how to debug?

[Gerben Wierda]

Solved. Or a workaround at least. By providing daemondo with an explicit pidfile (clean), daemondo keeps unbound alive, even if the first attempt at starting after boot fails to get the port. The second attempt (and any after that) succeeds.

Gerben Wierda
Chess and the Art of Enterprise Architecture <https://ea.rna.nl/the-book/>
Mastering ArchiMate <https://ea.rna.nl/the-book-edition-iii/>
Architecture for Real Enterprises <https://www.infoworld.com/blog/architecture-for-real-enterprises/> at InfoWorld
On Slippery Ice <https://eapj.org/on-slippery-ice/> at EAPJ

> On 16 Aug 2019, at 17:27, Gerben Wierda <gerben.wierda at rna.nl> wrote:
> 
> Additional info I was able to coax some error messages from launchd:
> 
> 2019-08-16 16:53:59 unbound: Starting process
> 2019-08-16 16:53:59 unbound: Running start-cmd /opt/local/etc/LaunchDaemons/org.macports.unbound/unbound.wrapper start
> Aug 16 16:54:00 unbound-albus[219:0] error: can't bind socket: Can't assign requested address for 192.168.2.66 port 53
> Aug 16 16:54:00 unbound-albus[219:0] fatal error: could not open ports
> 
> by changing the Startup call in the wrapper to
> 
>         (/opt/local/sbin/unbound-anchor -a /opt/local/etc/unbound/root.key 2>&1) || : && (chown unbound:unbound /opt/local/etc/unbound/root.key 2>&1) && (/opt/local/sbin/unbound 2>&1)
> 
> (stdout is faithfully reported by launchd, stderr is a black hole)
> 
> It seems the wrapper runs without proper authorisation (i.e. root) or the port has already been taken. The latter is not the case (no after boot anyway, during boot it is highly unlikely). Still investigating.
> 
> Gerben Wierda
> Chess and the Art of Enterprise Architecture <https://ea.rna.nl/the-book/>
> Mastering ArchiMate <https://ea.rna.nl/the-book-edition-iii/>
> Architecture for Real Enterprises <https://www.infoworld.com/blog/architecture-for-real-enterprises/> at InfoWorld
> On Slippery Ice <https://eapj.org/on-slippery-ice/> at EAPJ
> 
>> On 16 Aug 2019, at 12:04, Gerben Wierda <gerben.wierda at rna.nl <mailto:gerben.wierda at rna.nl>> wrote:
>> 
>> I have two ports active: unbound and an (updated by me, not yet committed) nsd.
>> 
>> Both have a startupitem that is identical, in that it uses the wrapper.
>> 
>> Both should launch at boot, but unbound doesn’t while nsd does. After boot, two daemondo instances are running, one for unbound and one for nsd. But only the one for nsd has actually spawned  nsd. The daemondo service for unbound has not launched an unbound instance.
>> 
>> When I do port load unbound after boot, the instance doesn’t load
>> When I do port reload unbound, the unbound instance loads without a problem
>> 
>> Apart from getting to the logging of unbound to look for clues, and increasing verbosity for daemondo in the startupitem, are there otherthings within macports I can do to debug?
>> 
>> Gerben Wierda
>> Chess and the Art of Enterprise Architecture <https://ea.rna.nl/the-book/>
>> Mastering ArchiMate <https://ea.rna.nl/the-book-edition-iii/>
>> Architecture for Real Enterprises <https://www.infoworld.com/blog/architecture-for-real-enterprises/> at InfoWorld
>> On Slippery Ice <https://eapj.org/on-slippery-ice/> at EAPJ
>> 
> 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macports.org/pipermail/macports-users/attachments/20190817/7411c608/attachment.html>