GitLab

Add Gluster to Ganeti by essentially cloning the shared file behaviour
everywhere in the code base.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

This method accepts a port number and checks that it is in fact valid.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

This will allow code reuse for Gluster through composition, rather
than inheritance.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Move the FileStorage class in its own file, together with its helper
functions.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

PathJoin fails with an unclear message if only one argument is passed
to it. Calling PathJoin("/foo") causes this exception:

    Error: path joining resulted in different prefix (/foo != /foo)

However, /foo and /foo obviously share prefixes: what this function
really checks is that the resulting path should be inside the first
argument and, in that sense, /foo is not a "real" prefix of /foo.

Given that PathJoin already asserts that it received any arguments
at all, this commit discards the assertion for an error to be raised
if 0 or 1 arguments are received. The semantics don't change, but
the error message is clearer.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

A large part of the complexity in this function is due to the need
to translate from "template-specific" parameter names to
"template-agnostic" parameter names. This logic is complex and having
complex code for complex logic is okay.

However, for all other cases, the additional complexity is not needed.
Additionally this function would do the wrong thing silently for new,
unhandled disk templates by just not changing the resulting paramters.

This commit replaces the ad-hoc logic with a more generally useful
logic while leaving the DRBD code carefully alone.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

This rather important dictionary from constants to classes was hiding
between function definitions. The dict cannot go to the top of the file
as the classes haven't been defined there yet, so it's been pushed
to the bottom of the file.

For bonus points, the dict gets an epydoc docstring and a comment, too.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Thomas Thrainer <thomasth@google.com>
Reviewed-by: Thomas Thrainer <thomasth@google.com>

Ganeti checks for orphan volume by making sure that it knows about all
volumes on disk; any additional orphan volume, even if created by the
administrator, causes a failure in gnt-cluster verify. Given that
hail and hbal already account and warn for missing space, this is
unnecessarily strict.

This commit demotes this error to a warning.

This commit also updates the part of QA about orphan and reserved
logical volumes to match, adapting the QA infrastructure slightly.

Fixes Issue 314.
Signed-off-by: Santi Raffa <rsanti@google.com>
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

As luxid now understands all the requests used by the command-line tools,
switch the default luxi socket for those to be the socket of luxid.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Support the query for the fields available for instances.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

...to be consistent with the python implementation.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

In commit 32933325

 this was only done for the Haskell side; do
so for python as well, to have both views consistent.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

When asked for all fields, we promise to return the list of fields
sorted according to niceSort. Keep this promise.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Luxid is more strict with closing the connection after receiving
a syntactically incorrect request, gnt-debug cannot use the same
client for several successive tests verifying that a syntactically
incorrect request is recognized as such.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

As the calling of watchFile and the evaluation of the initial
getFStatSafe takes non-zero time, the file could have changed
before inotify was set up properly. Solve this problem by an
additional check for the watched value to have changed immediately
after setup of inotify.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Currently they are generated only as Strings.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Jose A. Lopes <jabolopes@google.com>

This greatly enhances code readability.

Also fix monadic types "Q ExpQ" [which is "Q (Q Exp)"] to "Q Exp".
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Jose A. Lopes <jabolopes@google.com>

It's the same trick ShowS uses. We add a type class function for
showing a list to PyValue and then just use it in the instance for
`[a]`. This way we have the proper String instance without any
overlapping/incoherent instances.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Jose A. Lopes <jabolopes@google.com>

Now the file contains the type class declaration as well.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Jose A. Lopes <jabolopes@google.com>

This puts all PyValue code into one module, getting rid of orphan
instances.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Jose A. Lopes <jabolopes@google.com>

The time in SetWatcherPause is optional (with Nothing meaning
that the pause should be canceled), but the serialization is
not that of a Maybe Double; instead Just values serialize as
they are and Nothing serializes to null. Fortunately, we already
have a field modifier for this. So use it.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

Make luxid handle the QueryConfigValues call providing certain
simple status information about the cluster.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Add a predicate, in IO, to test whether the watcher is
paused.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Extend Path.hs to also provide the path to the file indicating
whether watcher is paused.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Make luxid handle SetWatcherPause correctly.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

With luxid taking over responsibility for handling watcher-pause requests,
it needs to know about this RPC. So have it available in Haskell as well.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

A JSON null value is used to indicate that the pause should be canceled.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

The instantiation of RPC requires a bidirectional functional dependency
between call type and return type. Hence we cannot use Unit everywhere.
Signed-off-by: Klaus Aehlig <aehlig@google.com>
Reviewed-by: Petr Pudlak <pudlak@google.com>

Change the IP of the instance to make it correspond to the one used in the
example.

The .253 is chosen so that, if needed, in the future the network can be changed
from a /24 to a /28.
Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

Add the document describing a new design for the OS installation process for
new instances.
Signed-off-by: Michele Tartara <mtartara@google.com>
Signed-off-by: Jose A. Lopes <jabolopes@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

No code is changed in this patch (except imports and qualifiers), only
moved.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

... to be usable for WConfd as well. A daemon handler is encapsulated
into `Handler` data type, which is then passed to a generic `listener`.

The changes are done in Luxi.hs so that the differences are visible and
will be moved into UDSServer.hs later.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This allows to use logging with the ReaderT monad transformer.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This separates logging from IO, allowing to create unit tests in future
for functions that use it.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Using MonadError is more correct than just "fail" on an arbitrary
monad, and more scalable when using monad type classes or monad stacks.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This allows it to be used with MonadError.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

.. and ResultT.

While at it, generalize also the MonadPlus instance of GenericResult and
add some Functor/Applicative instances.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Return the method (as any instance of JSON) and the arguments of a call.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

This simplifies code for closing such a socket.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>

Instead of passing a bare server socket around, we pass it encapsulated
in a data type together with parameters such as read/write timeouts.
Signed-off-by: Petr Pudlak <pudlak@google.com>
Reviewed-by: Klaus Aehlig <aehlig@google.com>