The HTTP server design doc will refer to the LUXI protocol,
so conflicts need to be avoided.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Stephen Shirley <diamond@google.com>

Docutils 0.4 doesn't provide the “Directive” class and Sphinx supplies
its own version which doesn't have the “assert_has_content” method.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This adds some recently added flags to the man-page of gnt-node command.
It also describes their purpose and handling.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

As per discussion on the man-page[1] update, this functionality should be
removed and replaced by just give the command to run if the user insists
of power cycle/power off the master and refuse to operate.

[1] http://groups.google.com/group/ganeti-devel/browse_thread/thread/95d4879a747cc295



This also removes the --all flag from gnt-node power because it was
buggy and would have added too much logic for special cases.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This doesn't cover everything yet, but it's a good demonstration on how
to use “pyeval” and “pyassert”.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

There are quite many hardcoded constants (e.g. “[…] one of ``file``,
``lvm-pv`` or ``lvm-vg`` […]). By using constants it'll be easier to
identify these.

With such lists of values it's also easy to miss some when
extending/changing something. By adding assertions inlined with reST,
these can also be detected.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Import with full name instead to reduce potential conflicts.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

If left in data, FillOpCode will complain. This fixes the QA breakage
after commit 526a662a.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

OpInstanceRename uses “instance_name” (like almost all other OpInstance*
opcodes), not “old_name”, to specify the original name. OpGroupRename is
made consistent by renaming “old_name” to “group_name”.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

- Use FillOpCode and unify parameter names between RAPI and opcode
- Generate parameter documentation
- Improve opcode parameter documentation

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

Some RAPI parameters don't match the name of the underlying
opcode. With this patch they can be renamed while filling
the opcode.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: René Nussbaumer <rn@google.com>

This fixes the doc builds broken by commit b292b986, but is only an
ugly workaround due to our dependency chain.

opcodes.py imports (both directly and indirectly) constants.py, which
imports _autoconf.py. Hence one could say that all .py files have a
logical dependency on _autoconf.py, since you can't "import opcodes"
before _autoconf.py is built. So the sphinx rule should depend on
_autoconf.py as a prerequisite.

But we can't use a normal prerequisite, since that would cause the
html docs to be rebuilt whenever _autoconf.py gets rebuilt, i.e. even
on end-users machines after ./configure; this is bad, and it also
breaks distcheck (as the html files are not removed in distclean,
since we do want them distributed).

So we only add an order-only prerequisite, which does the right thing
(although our dependency chain is still ugly).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Adeodato Simo <dato@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This replaces hardcoded lists of parameters with automatically generated
ones, making maintenance easier.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

For now this just registers a single new directive, “opcode_params”,
which can be used to generate opcode parameter documentation.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

These will be used in automatically generated parts of the RAPI
documentation. Not all opcodes are documented so far, but each added
opcode's parameters need to be fully documented (verified by an
assertion).

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

For now they're all set to None, but a following patch will set descriptions.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This patch adds text descriptions to all T* checks. Converting to a
string (e.g. str(ht.TNone)) will produce a user-readable text. This
will be used for Opcode parameter documentation.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Currently, when reinstalling an instance, especially when selecting an
OS interactively, is hard to be sure one has chosen the correct OS.

This patch adds displaying the selected OS. I'm not entirely happy, as
it's not very good English (e.g. keeping the same OS should be "but
keep the same OS", and changing should be "and also change …") but I
didn't find a better handling of all the cases:

node1# gnt-instance reinstall -o lenny-image+default instance1
This will reinstall the instance 'instance1' (and change the OS to
'lenny-image+default') which removes all data. Continue?
y/[n]/?: n

node1# gnt-instance reinstall instance1
This will reinstall the instance 'instance1' (and keep the same OS)
which removes all data. Continue?
y/[n]/?: n

node1# gnt-instance reinstall instance1 instance2
The reinstall will operate on 2 instances.
Note: this will remove *all* data for the below instances! It will
keep the same OS.
Do you want to continue?

node1# gnt-instance reinstall -o lenny-image+default instance1
instance2
The reinstall will operate on 2 instances.
Note: this will remove *all* data for the below instances! It will
change the OS to 'lenny-image+default'.
Do you want to continue?

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This allow to pass in --oob-timeout to define the timeout before the
out-of-band helper is beeing aborted.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This also has the sideeffect of presenting a table with the result

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

The tty can have strange side effect as it seems to replace \n with
\r\n. In case of simply catting a file and compare it's output this
will break. This patch adds the possibility to disable tty if needed.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

In commit bfceedbe a check was added to put the master at the end or
skip it completely. While this functionality works, it was done at the
wrong point because node_names was already processed to a node list
without this checks. The Exec method if the LU just used that node list.

To overcome this situation we correct the node_names before the node
list is generated.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This will allow us to use this functionality also in other cli-tools
like gnt-node power where we've to operate on multiple nodes.

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Signed-off-by: Adeodato Simo <dato@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

This was stating "secondary nodes" were the keys of the dictionary, when
they are primary nodes. Also, further clarify only the node's secondary
instances are included.

Signed-off-by: Adeodato Simo <dato@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

This patch adds ignore_status to ignore the offline flag of nodes
and also adds a force_master option to force operations on master node
if they will make the master unavailable (for some time).

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>

Signed-off-by: René Nussbaumer <rn@google.com>
Reviewed-by: Michael Hanselmann <hansmi@google.com>