Instance policies are changed and checked for changes.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The main parts, moved to qa_util, will be used for node groups too.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The old command-line options weren't working any more. Utility functions
are updated, and broken tests are removed (new tests are coming in the
following patches).

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Test upgrade and downgrade from a realistic 2.7 configuration.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

This makes easier to test real configurations.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Partial specs are filled with cluster defaults.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Command line options are brought in line with the specs change of previous
patch. Old options are still allowed in gnt-cluster init, where the
semantic will remain non-ambiguous even after introducing multiple specs.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Specs used to specify limits in instance policies are always complete, even
at group level (only the whole limit set can be overridden). This is in
preparation for introducing multiple limits.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

This makes the code more modular, in preparation for the following patches.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The tests check the interactions between the two commands.

Also refactored group QA code to separate tests about ipolicy, as those
tests are going to grow in complexity.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The output of this command can be used to create an exact copy of the
current instance policy specs. The command could be expanded to print all
the options used to create a group or the cluster.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

These options allow to specify whole instance policy specs. This is needed
for the upcoming changes that tend to threat specs as monolithic objects.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

This will be used for the new instance specs options.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Previously the string value was being corrupted.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

They were missing from the man pages.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The burnin uses its own option list, different from the one of all other
command line programs, and uses it as a parameter to a function that now
requires the "--reason" option to be present.

This commit updates the burnin option list to be compatible with the reason
trail implementation.

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

In a simulated cluster as created by the simulation backend to
the htools, make the first node of the first node group the master
node. In this way, hools (like hroller) that require a master node
can also be used on simulated clusters, e.g., for testing.

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

Add a unit test for hroller in a situation where the order
is completely determined. There are 3 nodes where node 2 has,
for each of the other nodes, an instanced shared with it. So
node 2 has to be a reboot group of its own. There are no more
instances, hence, my minimality, the other two nodes have to
form a single group. Node 1 is the master, and hence has to
be rebooted last; this determines the order.

Also add a unit test verifying that hroller rejects configurations
with multiple master nodes, even if the option --force is given.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Sometimes, e.g., for hroller, it is necessary to know which node
is the master node. Therefore this information has to be included
in the text format as well. Since we never use an offline node as
master node, we can put this information can be put in the "is
offline" field; as old versions only check this field for equality
to "Y", using "M" to indicate the master is compatible in both
directions. This also leaves room to add more roles in the future,
should we need them.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Signed-off-by: Sebastian Gebhard <sege@fs.ei.tum.de>
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This commits allows ganeti to correctly forward the reason trail information
regarding instance reboot.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Also, this add an infrastructure for having parameters common to all the
RAPI calls.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Some parameters can be common to all the RAPI calls: this commit
adds the possibility of specifying them just once, instead of
manually adding them to the classes describing each call.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The reason trail will contain an item indicating the job_id and the
index number of the current opcode inside the job queue.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The reason trail is available for all the opcodes, and as such it is
initialized as a generic option.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The reason trail will be available for all the OpCodes, so the field that
is to contain it has to be added to the OpCode base class.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Noded functions will have to extend the reason trail specifying what
operation is being executed.

This commit adds a function for doing that (will be used by future commits).

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Add a function to convert an opcode class name to a source string for the
reason trail, factoring out the code in common with the existing _NameToId
function.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

We are interested in storing the reason trail for operations that changed
the state of instances. This commit adds a function to do this.

Also, the GetInstReasonFilename() function name is prepended with an
underscore because it is only used in this file.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The timestamp is returned as an integer number of nanoseconds since the Unix
epoch.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

This will be used as the type for the reason trail parameter.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

A few constants are required, describing the source reason of the opcode.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Remove the useless parts of the old, partial, implementation of the support for
tracking the reason of instances state change, before implementing the new
reason trail support, as per the design document.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Adding the reason trail as a parameter instead of as a new class attribute fits
better with the current structure of the system.

Also, the description of what part of the implementation will be OpCode
specific and which will be shared is added.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

The script was added to Ganeti before the monitoring daemon, so it
did not include that user yet.

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

In commit 3dd990, the possibility of taking a default reviewer
from the environment was introduced. Fix two deficits here
- perl does not inherit the environment
- add at all places.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This is similar to 405692ef, reverted in aeefb4cd.

Signed-off-by: Bernardo Dal Seno <bdalseno@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

As people rely on the master node being the last node of the last
group, make hroller fail, if no master node could be found in the
cluster. This happens, e.g., if a backend format is used that does not
contain information about the master node. The error can be turned
into a warning by using the --force option.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

Where the information is available, hroller now schedules the
master node as the last node in the last group. Update the man
page to reflect this change.

Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>