GitLab

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

In tiered allocation, if by shrinking only a single resource a valid
allocation can be found, shrinking is bound to shrink on this resource.
Of course, after shrinking that resource a little bit without finding
an allocation, this property is still valid. So we can as well shrink
on that resource as far as needed to get a valid allocation.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

* stable-2.7:
  Provide the right fix for the kvm_extra parameter
  NEWS and version updates for 2.7.0~rc3
  kvm: don't pass whole extra as a single argument
  For node queries allow short forms of host names
  Provide a special filter for host names

Conflicts:
	NEWS: trivial
	configure.ac: trivial
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

With the previous adjustment the arguments were boxed into one extra
array. This actually makes the parameter work (with the same caveats).

The manpage is also updated with the limitations.
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Time for a new release.
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

If extra is longer than a single word passing it the way it was passed
won't work. Of course this solution is also only partial, as won't take
into account any form of quoting. :/
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

For node queries use the host-name filter instead of the simple
equality-based one.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

For host names, usually short forms are used, e.g., node1 or node1.sub
instead of the full qualified node1.sub.example.com. Therefore comparing
node names only by equality is too restrictive. This patch provides an
alternative makeHostnameFilter to makeSimpleFilter suitable for comparing
host names.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This fixes issue 485. In the python implementation of
the cluster config info, the enabled_disk_templates
were missing.
Signed-off-by: Helga Velroyen <helgav@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Some of the documents have the version number of Ganeti. This commit updates
it in preparation for the 2.8 version bump.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

The recently introduced check for python libraries required only for testing
(commit 27df5b73

) was not correct. This
patch fixes the issue.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

* stable-2.7:
  Add a warning about the restriction of gnt-node evacuate
  Fix permissions of the queue archive dir
  Fix daemon QA tests
  grow disk: call SetDiskID before blockdev_getsize
  Catch DeviceCreationError unhandled exceptions

Conflicts:
	lib/cmdlib.py (due to the cmdlib split)
	qa/qa_daemon.py (trivial)
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

_CreateBlockDevInner is called twice in TLReplaceDisks. While this
should be fixed, right now this leaves the DeviceCreationError exception
unhandled, which causes a problem due to the fact that this exception
was never supposed to be sent over the wire. For 2.7 we just catch the
exception and convert it to an OpExecError, but we should later do a
more proper cleanup.

This addresses Issue 472.
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Missing this call caused blockdev_getsize to be called on the wrong
physical device, and fail.

This fixes Issue 497.
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

TestInstanceConsecutiveFailures now leaves the test instance in the same
state it got it.
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Some python libraries are only needed for the tests, but not for running
Ganeti. If those libraries are missing, just disable the Python tests.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Add command for updating the list of cabal packages, before starting
installing them.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This patch fixes some typos in the documentation index.
Signed-off-by: Weiwei Jia <harryxiyou@gmail.com>
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

The fact that confd has been given access to the server.pem certificate
needs to be described in the security document.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

The permissions were set to 0740, but of course being a directory the
correct permissions are 0750.
Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Even though gnt-node evacuate should be able to evacuate all instances
at once, this currently is not the case. Add a warning about this deficit
for the time being.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

The chroot only had tools for building and testing Ganeti, but not for
building its Debian packages. They are added by this commit.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Give more precise instructions on the next steps to actually run the
chroot.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

The configuration file are included in build_chroot, and are
automatically created by the script if it is not able to find them
already in the directory pointed by $DATA_DIR.

Note that the configuration file directory is cleaned if it's temporary,
but it is left untouched if it was esplicitly specified by the user.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

The server.pem file is 0440 (materd_uid, masterd_gid) and cannot be
accessed by confd.

This patch contains a workaround for 2.8.0~beta1 (adding conf to the
masterd group in the user setup tool, and remembering the users to
actually execute that tool while upgrading).

This partially fixes Issue 292.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

The heuristics for tiered allocation has been improved in that it
chooses to shrink such a resource next where by shrinking only this
resource a valid allocation can be made, if such a resource exists.
In order to decide whether such a resource exists, all potential
sizes of this resource are verified. We can improve performance by
cutting, as soon a potential allocation of an instance is no longer
blocked on this resource.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

* stable-2.7
  More agressively line break man pages
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Newer versions of pandoc (as, e.g., shipped with Ubuntu 13.04)
more agressively quote, in particular minus signs. That sometimes
leads to unintended line breaks, which are a problem, if the new
line starts with a dot.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Now that alpha versions are supported, we can bump the version number
for branch stable-2.8 to 2.8.

This also requires updating the cfgupgrade tool.

Thanks to the previous patches, all the other documents can be upgraded
during the alpha lifetime, before switching to beta (that will enable
the version number checks).
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

One of the unit tests checks whether cfgupgrade's downgrade option is
idempotent. It is, but a version number check made it impossible to
actually downgrade multiple times.

With this change, both the current-version version number and the
target-version version number are accepted by the downgrade tool.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

The downgrade option of the cfgupgrade tool was not changing the version
number in the configuration file of the cluster while performing the
downgrade.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

In this example, memory prevents adding more than one instance on
the first node. Yet, on the other nodes, 2 instances can be placed
each, if disks are shrunk appropriately. This, however, requires
to courageous decision to shrink disks next, even though more nodes
fail to allocate more due to memory than to disk.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

In tired allocation, hspace shrinks that resource of the instance
next, that causes failure on most nodes. While, this is not a bad
strategy in general, it can lead hspace into a dead end if for a large
number of nodes a particular resource blocks any further allocation of
policy compliant instances. So we improve the heuristics in that it
chooses to shrink such a resource next where by shrinking only this
resource a valid allocation can be made, if such a resource
exists. This improves the results in some cases, while still keeping
the computational complexity of the algorithm low.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

For a function f :: a -> GenericResult a, iterate it (in the sense of the
monad), until the result is Bad; return the list of values occurred.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

GenericResult, while rightfully a type of its own, is isomorphic
to Either. So, also provide the case analysis function (i.e., the
universal arrow out of the sum).
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Alpha versions should not check for version numbers in READMEs
and documentation.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Alpha versions are still under heavy development, and can therefore not have an
entry in the NEWS file yet. The entry will have to be added before switching to
the first beta version.

Partially fixes Issue 448
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>