pcp
[Top] [All Lists]

Re: pcp updates - assorted non-trivial changes

To: "Frank Ch. Eigler" <fche@xxxxxxxxxx>
Subject: Re: pcp updates - assorted non-trivial changes
From: Ken McDonell <kenj@xxxxxxxxxxxxxxxx>
Date: Fri, 10 May 2013 09:55:46 +1000
Cc: pcp@xxxxxxxxxxx
Delivered-to: pcp@xxxxxxxxxxx
In-reply-to: <20130509115319.GB19133@xxxxxxxxxx>
References: <518ABC53.2090207@xxxxxxxxxxxxxxxx> <y0ma9o55bs8.fsf@xxxxxxxx> <518B087D.702@xxxxxxxxxxxxxxxx> <20130509024318.GA19133@xxxxxxxxxx> <518B1217.2020307@xxxxxxxxxxxxxxxx> <20130509115319.GB19133@xxxxxxxxxx>
User-agent: Mozilla/5.0 (X11; Linux i686; rv:17.0) Gecko/20130329 Thunderbird/17.0.5
On 09/05/13 21:53, Frank Ch. Eigler wrote:
...
That seems overly invasive for exposing data on private people running
private pmloggers for their private purposes.  Can't this be opt-in,
or depend on userid? ...

It could be, but that was not in scope 10 years ago when this was designed and built ... 8^)>

When we get secure and authenticated services working for client-pmcd
connections we probably should visit the pmlc-pmlogger connections as well, although the number of people using pmlc is (I am pretty sure) very small.

... Is it even safe for a system daemon to trust &
rebroadcast data from some random unprivileged pmlogger?

The data from the pmlogger that is available from pmcd is minimalist ... see pminfo -T pmcd.pmlogger ... I doubt that there is much cause for concern there.

Note that all of these issues need to be repeated for pmie and the pmcd.pmie.* metrics.

...
It sounds like the debian one-time post-install scripts would be the
place to set permissions that are not-quite-right after untarring,
rather than the system reboot-time rc files.

Agreed, but we're back to QA effort and risk for a 3.8.0 release.

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