[bind10-dev] component shutdown should handle hanging process

[Michal 'vorner' Vaner]

Hello

On Fri, Oct 26, 2012 at 01:32:40PM +0200, Shane Kerr wrote:
> > • The <component> shutdown should really be hidden. This is internal
> > command and is not expected to by called by user (partly because it
> > doesn't do what one would expect, since the component gets restarted
> > immediately).
> 
> True, we have an open ticket for that:
> 
> http://bind10.isc.org/ticket/789
> 
> We can put on next-sprint-proposed if it seems worthy of fixing now.

I don't know. Probably after we first the second point first. It is kind of
useful now.

> > • There's no way to „disable“ a component in boss. It can be only
> > removed from the configuration, with all the settings on how to start
> > it.
> 
> See also this ticket:
> 
> http://bind10.isc.org/ticket/795
> 
> I agree we should have a way to disable a component. Does this already
> exist as a ticket?

I don't know, I think this is the exact ticket. It doesn't really have to be a
command (except for the restart maybe), it can be modification of config:

config set Boss/components/b10-something/enabled False
config commit

> > • If the boss wants to shut down a component and it is stuck, it just
> > fails to do so and gets stuck in the state where it is trying to shut
> > it down for ever.
> 
> Probably we should use the same logic here as we use when the system is
> shutting down.

+1 (though it probably needs a deeper change in the code, the current state is
the first proof-of-concept implementation O:-).

> > Of course, we should never have anything stuck in the first place O:-)
> 
> That's why we use C++, because we choose simply never to make any
> mistakes. ;)

It reminds me of a friend (and a programming guru). When asked how he debugs, he
usually answers something like „Well, because debugging takes so much time, I'm
usually very careful not to write the bugs in the first place. And even if I
write one, I usually run the program, compare the output with the correct one
and think about how exactly would it have to be wrong to produce exactly this
wrong output“ :-)).

With regards

-- 
When a fly lands on the ceiling, does it do a half roll or a half loop?

Michal 'vorner' Vaner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <https://lists.isc.org/pipermail/bind10-dev/attachments/20121026/c9aa4525/attachment.bin>