pcp
[Top] [All Lists]

Re: [pcp] pcp updates: qa, containers

To: Nathan Scott <nathans@xxxxxxxxxx>
Subject: Re: [pcp] pcp updates: qa, containers
From: "Frank Ch. Eigler" <fche@xxxxxxxxxx>
Date: Wed, 18 Mar 2015 10:38:21 -0400
Cc: Mark Goodwin <mgoodwin@xxxxxxxxxx>, pcp <pcp@xxxxxxxxxxx>
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <1761824975.8325622.1426651846865.JavaMail.zimbra@xxxxxxxxxx>
References: <839234551.7808463.1426575003930.JavaMail.zimbra@xxxxxxxxxx> <1452631543.7808485.1426575023285.JavaMail.zimbra@xxxxxxxxxx> <y0m7fufmqxd.fsf@xxxxxxxx> <5508CEA0.3060107@xxxxxxxxxx> <1761824975.8325622.1426651846865.JavaMail.zimbra@xxxxxxxxxx>
User-agent: Mutt/1.4.2.2i
Hi -

On Wed, Mar 18, 2015 at 12:10:46AM -0400, Nathan Scott wrote:

> [...]
> > I don't know how yet, but PCP PMDAs (other than those that are default
> > pre-installed) are probably going to need meta-data if they end up running
> > in separate containers[...]
> 
> Hmm, I'm not sure there's alot of win to be had from breaking out the default
> PMDAs - will end up with alot of packages/containers that everyone needs just
> to get started.

Indeed, that seems to be going overboard.


> > > By the way, have y'all considered using the daemons in foreground mode
> > > ("pmcd -f" et al.) as the contained pid-1 process, instead of
> 
> FWIW, this was the way it started out until we hit several problems with
> that simpler approach.

Can you elaborate (or link to prior findings)?


> > > container pid-1 would actually exit when the daemon does, [...]
> 
> Not sure what code you're looking at, but that *is* what happens already.

If I read the current code correctly, it'll interpret any transitive
child process's death as a signal to exit, not specifically pmcd.  If
any daemonized children are started within the container, their death
could be routed to "pmpause" and cause an early death.


- FChE

<Prev in Thread] Current Thread [Next in Thread>