Many of our functions have to follow a given API, and thus we have to
keep a given signature, but pylint doesn't understand this. Therefore,
we silence this warning.

The patch does a few other cleanups.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

The node_list argument to _DiagnoseByOS is not used, and is obsoleted by
the fact that the rlist argument already has the valid nodes as keys
(assuming RPC behaviour didn't change). Thus, we remove it and silence
the warning.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

The disk template is not needed, all that's used is the disk data. As
such, remove this parameter from the function.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Currently only the rpc call, but not its description (which also shows
the argument) is logged. We change this to log failmsg too, and this
also silences a warning.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

So we just silence the warning.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

The 'name' argument is not used anymore, probably since before 2.0.
Since this is an internal function, we can just remove it (from its
caller too).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Plus a silence for a wrong "uninitialized var".

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Many methods are simple pure functions, and not depending on the object
state. We convert these to staticmethods.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This patch should have only:

- pylint disables
- docstring changes
- whitespace changes

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

The chroot and fake hypervisors were missing:

- the powercycle node functionality
- proper handling of migration requests

The powercycle was just used as in the other hypervisors (use the
standard linux powercycle). The migration for chroot was disabled
explicitly, whereas for the fake one it was implemented to simulate
correctly. This required some work on the fake hypervisor, but now the
implementation of start/stop/etc. is much more clean.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This patch adds explicit errors (instead of notimplemented) in
FileStorage (and the associated TODOs).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Wrong argument name and wrong number of arguments in string formatting.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This patch abstract the check "is instance stopped" into a separate
function, and thus simplifies a couple of higher-level functions. It
also moves from manual read of the pidfile to use the (correct
abstraction of) _InstancePidAlive.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

In some cases we only need the pidfile, but not the pid or the alive
status.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Detected by an 'Unused variable' warning.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Note there are some cases left which need extra cleanup.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

The logging functions expand the arguments themselves, thus it's safer
to let them do it rather than manual string formatting.

Also re-wraps one comment.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

Per pylint warnings.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This just adds a stub function that raises an assertion error; this
accomplishes two things:

- silences many pylint warnings
- if we ever stumble upon this, a specific assertion error is
  (hopefully) clearer than just a not implemented error

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This removes/updates some module-wide pylint disables.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

New version of pylint, new bugs found!

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This patch adds targeted pylint disables, where it makes sense (either
due to limitations in pylint or due to historical usage), and also a few
blanket ones in rapi where all the names are… “different”.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

This cherry-picks the utils.FieldSet.Matches changes and the significant
jqueue.py change. These are stable in the 2.1 branch and therefore make
sense to backport to 2.0 (are basically cleanups).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Olivier Tharan <olive@google.com>

MaybeRaise in lib/errors.py had a typo in the doc string

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

Since gnt-instance start doesn't do any checks on the IP, it doesn't
make much sense to do so in instance create (with start) if the user
expressly passes in ‘--no-ip-check’. Removing this requirement eases the
no-name-check mode (otherwise one would have to create --no-start and
then only start).

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This patch adds --no-name-check to gnt-instance add and gnt-backup
import. This is opposite to the opcode parameter (name_check) as it is
similar to ip_check and start.

It also adds it to RAPI and gnt-instance batch-create as a parameter in
the input (JSON-formatted) file.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This adds a new opcode parameter ‘name_check’ (similar to ip_check) that
is not required to be present (to easy backwards compatibility for
tools).

It also adds a CheckArguments to LUCreateInstance and changes the
workflow related to instance IP checks and NIC initialisation based on
it.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

The cluster verify checks for fqdn are done via address lookups, and
there we actually use the FQDN. However, for the ssh hostname check
which is done at node add time, we rely on the default of the “hostname”
command. And Debian for example recently changed the default to return
the shortname unless one passes ‘--fqdn’.

This patch is imported from the Debian packaging.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This patch unifies the search for external script to always go through
utils.FindFile and implements in that function a restriction on valid
chars in file names and (additionally) that the passed name is the
basename of the final (absolute) name.

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

This will allow reuse of the same mask for multiple validations.

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

In most uses of LUQueryNodes, we don't take a lock. This means that the
instance data is not protected across GetInstanceList and
GetInstanceInfo, and this can lead to instances not existing anymore.

Switching to GetAllInstanceInfo means that we get a single,
semi-consistent snapshot (since instances can still be modified, but
it's much better).

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

This patch adds the 'cache' parameter for KVM; currently this is only
customisable at the hypervisor level, so it's the same for all drives
(except any CDROM image, which gets the default).

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

On some distributions pyinotify is installed in a different way, and the
actual module just contains an internal pyinotify entry, which is the
actual library. On others the main pyinotify module contains the library
itself. We'll try both, in order to be more compatible.

Signed-off-by: Guido Trotter <ultrotter@google.com>

By allowing also the primary ip field to be fetched directly, we avoid
one more confd lookup, or dns request, to find out which address the
master node lives at.

Signed-off-by: Guido Trotter <ultrotter@google.com>

Change the ClusterMasterQuery to allow a query, and if present accept a
list of fields to be returned. Currently only name and ip are accepted.

This feature will be used by NLD to route the cluster ip over the nbma.

Backwards compatibility is preserved.

Signed-off-by: Guido Trotter <ultrotter@google.com>

The documentation for file objects' read method states that if the size
is "negative or ommitted", all data is read; thus we can simplify it to
have size=-1 by default and not have the if test.

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

The DRBD driver can ignore dead disks but not dead meta devices (for
which it refuses to configure the minor). To handle this case, we check
that the meta device is readable and if not we ignore it (the same as
when backend._RecursiveAssembleBD didn't find it).

A needed change is the move of some checks and initialisers before this
test (which is before the super().__init__ call), but that should be
fine.

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

The warning will be generated if the clocks diverge by more
than 150 seconds. Due to the way the RPC system works, we
cannot get exact time differences, e.g. if one of the
queried nodes is broken. The comparision is done using a
time window.

Confd queries will fail if the clock on the client and server
are more than 300 seconds from each other. This check helps
keeping at least the nodes of a cluster in sync.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Guido Trotter <ultrotter@google.com>

This shouldn't happen, but if it does it's better to fail at this level,
rather than create a broken NIC script, which is hard to debug.

Signed-off-by: Guido Trotter <ultrotter@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>