I'm posting here in the hope that someone cares about PCP on the SUSE
distributions (sgi?).
I loaded the recently released openSUSE 12.1 on one of my VMs and found
the following "issues"
1. The package names for the libraries have been changed from
pcp-libs and pcp-libs-devel to libpcp3 and libpcp-devel
respectively ... besides being internally inconsistent (should
be libpcp3-devel me thinks), the more serious problems are
A. whoever did this did not send the patches up stream to
the PCP maintainers, and (more importantly)
B. there are no apparent Obsoletes rules to deal with
upgrades, so installing PCP 3.5.9 from oss.sgi.com
source on top of PCP 3.5.8 from openSUSE leads to a
botched install and parts of both 3.5.8 and 3.5.9
installed
2. The init.d framework has changed (again!), and now attempts to
run /etc/init.d/pcp produce "redirecting to systemctl" as the
only message on stdout or stderr and no further indication if
this worked, did nothing or failed (without dredging in the tail
of /var/log/messages)
3. /etc/init.d/pcp start has become a no-op if the pcp "service" is
already running ... you need either an explicit stop and start,
or a restart to really restart pmcd
Together 2. and 3. break a significant fraction of ALL of the PCP QA.
Hopefully, someone who knows more than I can point me in the direction
of reconfiguring bits of openSUSE 12.1 to return to a more sane state of
affairs.
I have a patch for 1. that moves the RPM names closer to the dpkg names
and handles the upgrades with package renaming issues ... I have no idea
how we can check if this passes the RPM police for the assorted
RPM-based distros (which I'll bet are not singing from the same style
guide).
But, wait there is more ... QA 067 KILLS the system ... the last output
is "redirecting to systemctl" then clunk, no kernel oops but absolutely
nothing works until you power cycle the VM.
Anything by way of suggestions to address some of these issues or lower
my blood pressure would be much appreciated.
|