I'd also like to see a command for starting additional copies of a homogeneous program config. In Chris' example, if the program frobnozz had numprocs > 1, I'd like to be able to issue a command like
addCopy frobnozz, effectively incrementing numprocs by 1. On Wed, Jul 23, 2008 at 7:52 AM, Chris McDonough <[EMAIL PROTECTED]> wrote: > Anders Quist wrote: > > On Fri, 18 Jul 2008 12:15:04 -0400, Chris McDonough <[EMAIL PROTECTED]> > writes: > > > >>>> This feature (reload a process) is much-asked for and would be very > >>>> useful. I can't speak for the other developers on the list, but I've > >>>> personally more or less been waiting to have that particular itch > >>>> myself or find someone who has that itch to fund the feature. > >>>> Twiddler is a good solution in the meantime, as Roger said, however > >>> Will you consider a patch if we submit one? > >> Of course! FTR, we're a bit sticky on tests, though. We don't really > >> accept patches that break the current test suite, or which add new > >> features that aren't tested. > > > > Some design issues: > > > > - How to decide if a proccess has had its configuration changed? I > > would prefer ProcessConfig objects to grow an __eq__ method, but it > > would be a hassle to know which attributes to look at, unless one > > does dir(self). > > What would __eq__ compare it against? > > For the record, rather than going for the "do what I mean" of comparing > running > process configs against the config file right away, a first step towards > this > goal would be to add the following commands to supervisorctl and > accompanying > backend handlers: > > add -- > > Add a process to the running config from a section in the config file. > You may not add a program that is already in the running configuration. > > remove -- > > Remove a process from the running config. The process must first be > stopped. > > That would be useful by itself. If you knew that you had just change the > frobnozz program config, you would issue: > > stop frobnozz > remove frobnozz > add frobnozz > > We'd implement more magical commands in terms of these. > > > - I don't see any way to do better than just restart a procces that > > has had _any_ change to its config. > > > > > - If a group has had a change to its config, should all processes in > > the group be restarted? Particularly, if processes are added to or > > removed from the group, should the whole group be restarted? Or just > > all processes with later priority? > > I think it'd be more useful to be able to do these things manually for now > in > terms of add and remove. > > - C > _______________________________________________ > Supervisor-users mailing list > [email protected] > http://lists.supervisord.org/mailman/listinfo/supervisor-users >
_______________________________________________ Supervisor-users mailing list [email protected] http://lists.supervisord.org/mailman/listinfo/supervisor-users
