Re: Making start/stop idempotent

[Mark Nottingham]

I realised I never sent the patch without the option;

idempotent_start.patch
Description: Binary data

On 12/08/2008, at 4:14 PM, Amos Jeffries wrote:

So, effectively just change to exit(0) in those two places?

to keep other users going.

Amos

For the HEAD commit(s) yes. If you need it ported the option may be needed

On 12/08/2008, at 4:02 PM, Amos Jeffries wrote:

Our management tools like processes to be idempotent; i.e., you
should
be able to start or stop a process any number of times without it
throwing an error.


(upon start) and when there isn't one (upon -k shutdown).


way to do it seems to be with a command-line option; I've somewhat
arbitrarily chosen -p.

Does this seem reasonable? If so, I'll submit a patch shortly.

Currently, Squid will return 1 if a squid process is already running I'm writing a patch to change this behaviour, and the most reasonable

I agree with Robert. The behavior change is nice. Requiring another
option
to enable/disable it is not.

I'd personally go with just the behavior change to the HEAD trees.
Only disabling and adding the option to enable if its ported to
currently
stable releases (2.7, 3.0).

Amos

--
Mark Nottingham       [EMAIL PROTECTED]

--
Mark Nottingham       [EMAIL PROTECTED]