On Sunday 08 February 2009, Udo Richter wrote:
On 08.02.2009 10:19, Ville Skyttä wrote:
On Saturday 07 February 2009, Udo Richter wrote:
Unfortunately it's not that easy. Currently, VDR backgrounds the call to the shutdown script, and detaches the shutdown script from the VDR process. Only because of that, the script can 'survive' the kill of VDR,
Why is that even necessary? Shutdown scripts could selectively background+detach things that need to survive killing of VDR and the script themselves, if any.
Detaching as a script isn't that easy, the reason for these ugly echo "..." | at now workarounds.
There are other alternatives to choose from if one for some reason can't stomach the "at" approach.
and only because of that the script can display messages via SVDRP.
Hmm, why wouldn't a non-background, non-detached script called by VDR be able to do that?
The main VDR thread would have to wait for the return, and SVDRP won't be handled until then. SVDRP connections from the script would timeout. You can see that if you use SVDRP commands from commands.conf without explicitly backgrounding.
That sounds like a bug to me, irrespective of this discussion.
The question for VDR is: If we don't shut down, when should VDR check again, or how does VDR know that the background job is done?
Why would VDR check or care in the first place?
An error level return is quite limited here.
Sure, but quite a bit better than what's currently sanely doable.