Re systemd.
Commit a59111b in my upstream tree turns on use of systemd in PCP for
all Debian-based platforms that configure thinks they're running a real
systemd. This code has been heavily exercised on the non-Debian
platforms for a long time, and my testing on Ubuntu 17.04 and 17.10
suggests
With the upstream changes mentioned above and a couple of tweaks
(everything up to and including commit f09cac4), I've been able to
complete a build, package, install and start PCP operation on all the
machines in my PCP QA Farm.
This means:
amd64 FreeBSD 10.2-RELEASE, amd64 OpenBSD 5.8
i386
I agree with Eric.
Between when this bug was opened and now we've done a lot of work on
hardening PCP.
There is a very good chance (based on the fact that I wrote much of the
related code) that this issue with pmcd was related to PDU handling
between pmcd and clients of pmcd. There have been
Update.
We've written replacements for system() and popen() that use execvp()
and do not call /bin/sh -c "some command". These are __pmProcessExec()
and __pmProcessPipe() in libpcp.
All uses of system() and popen() in the core libpcp library have been
converted to the new routines.
There are
On 29/08/17 10:44, Eric Desrochers wrote:
> @Dariusz, Seth, Ken, Frank and anyone else involved in this discussion.
>
> What would you guys think if we schedule a hangout call between
> Canonical & PCP upstream folks to discuss about all this ?
Sounds good to me ... we may be a little timezone
Apropos the CVE's all of these have been completely addressed in past PCP
releases.
CVE-2012-5530 - fixed in PCP 3.6.10 (released 19 Nov 2012)
CVE-2012-3421, CVE-2012-3420, CVE-2012-3419 and CVE-2012-3418 - all fixed in
PCP 3.6.5 (released 16 Aug 2012)
CVE-2001-0823 - fixed in PCP 2.2.1
+1 for getting a fix please (like you're competitor distros have been
able to do).
This is killing PCP (www.pcp.io) QA on all Ubuntu 16.04 platforms.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
