[Linux-cluster] Reporting GFS2 bugs

[Steven Whitehouse]

Hi,

Just a quick clarification on the bast way to report bugs and problems
relating to GFS2.

Firstly, if you are a paying Red Hat customer, all issues should be
reported via the Red Hat support team. That doesn't prevent you from
monitoring any bugzilla opened as part of that process directly, if you
want to. The information below applies to those who are _not_ paying
customers (i.e. members of the community)

All bugs (or potential bugs) should be reported via the Red Hat bugzilla
http://bugzilla.redhat.com and the appropriate distro selected. If you
are using CentOS, report it against the relevant RHEL version and
clearly mark the description of the bug as being from CentOS. If you are
using a non-Red Hat distro or an upstream kernel, then please report it
under Fedora/rawhide and again clearly mark the description with the
exact versions of the kernel and userland in question.

The reason that we ask for things to be reported in this way is so that
we have a single place where all the GFS2 issues can be queried. That
means both kernel and userland issues. Also it will automatically appear
on my todo list, it will be available for others to reference in case
they have the same or a similar issue, and for us to query in order to
look for patterns which may help us track down a problem faster.

Don't worry about whether your issue is a duplicate too much. If it is
obviously a duplicate, then please use the existing open bug, if you are
not sure, then report it as a new issue and we'll mark it as a duplicate
if required. Sometimes it isn't always obvious whether a bug is a
duplicate of an existing issue until the issue is found and fixed.

Bugs reported via IRC or mailing lists are tricky to keep track of, and
are difficult for other people to query in order to see whether they
have the same issue. By all means use those methods to ask about whether
a particular item is a bug, but please always file a bugzilla if that
turns out to be the case.

Questions sent to the mailing list are better since they are more likely
to be easily available via the archives for future reference by others.

One thing which is often missing from bug reports, but is usually vital
when working out what went wrong, is a basic description of the cluster
and the workload in question: how many nodes? What kind of storage? What
was the application which was running when the problem occurred? just
enough to give a picture of the basic architecture involved.

Don't just restrict bug reports to the code itself. If there are errors
in documents or things on the web site, or whatever, then we'd like to
know about those too.

Finally, don't forget that bugzilla can also be used to request new
features. In that case open your bug against fedora/rawhide and put
[RFE] in the description, and "FutureFeature" in the key words.

Do let us know if you have any questions or comments,

Steve.