Reported by Iustin.

Reviewed-by: iustinp

By design if an empty list of locks is acquired from a set, no locks are
acquired, and thus release() cannot be called on the set. On the other
hand if None is passed instead of the list, the whole set is acquired,
and must later be released. When acquiring whole empty sets, a release
must happen too, because the set-lock is acquired.

Since we used to overwrite the required locks (needed_locks) with the
acquired ones, we weren't able to distinguish the two cases (empty list
of locks required, and all locks required, but an empty list returned
because the set is empty). Valid solutions include:
  (1) forbidding the acquire of empty lists of locks
  (2) skipping the acquire/release on empty lists of locks
  (3) separating the to-acquire and the acquired list

This patch implements the third approach, and thus LUs will find
acquired locks in the acquired_locks dict, rather than in needed_locks.
The LUs which used this feature before have been updated. This makes it
easier because it doesn't force LUs to do more checks on corner cases,
which are easily forgettable (1) and allows more flexibility if we want
LUs to release (part-of) the locks (which is still a possibly scary
operation, but anyway). This easily combines with (2) should we choose
to implement it.

Reviewed-by: imsnah

A job should only exist once in memory. After the cache is cleaned,
there can still be references to a job somewhere else. If there
are multiple instances, one can get updated while a function is
waiting for changes on another instance. By using
weakref.WeakValueDictionary, which automatically removes instances as
soon as there are no strong references to it anymore, we can solve
this problem.

Reviewed-by: iustinp

Reviewed-by: ultrotter

It can be confusing otherwise.

Reviewed-by: ultrotter

This is a result of the log timestamp changes.

Reviewed-by: imsnah

Seems noone ran a burnin lately :)

Reviwed-by: amischenko,ultrotter

This is a large patch, but I can't figure out how to split it without
breaking stuff. The old way of getting messages by always getting the
last one didn't bring all messages to the client if they were added
too fast, thereby making commands like “gnt-cluster verify” less than
useful. These changes now introduce some sort a serial number per
log entry to keep track what message a client already received. They
also remove the log lock per opcode to make reading log entries thread
safe.

Reviewed-by: ultrotter

This gives continous output instead it being buffered.

Reviewed-by: ultrotter

Currently it can only be enabled by modifying utils.py, but we can
add a command line parameter later if needed.

Reviewed-by: schreiberal

Reviewed-by: ultrotter

I forgot to remove these when converting the QA configuration from YAML
to JSON.

Reviewed-by: ultrotter

Vim doesn't recognize the format automatically.

Reviewed-by: ultrotter

As for LUQueryInstances the first version just acquires a shared lock on all
nodes. In the future further optimizations are possible, as outlined by
comments in the code.

Reviewed-by: imsnah

This first version acquires a shared lock on all requested instances and
their nodes. In the future it can be improved by acquiring less locks if
no dynamic fields have been asked, and/or by locking just primary nodes.

Reviewed-by: imsnah

A few more tests written while bug-hunting. One of them shows a real
issue, at last. :)

Reviewed-by: imsnah

I was hunting for a bug in my code and thought the culprit was in the
locking library, so I added a test to check. Unfortunately turns out it
wasn't. :( Committing the test anyway, while still trying to figure out
what's wrong...

Reviewed-by: imsnah

If a list with a duplicate value is passed to a lockset what the code
now does is to try to acquire the lock twice, generating a
double-acquire exception in the SharedLock code. This is definitely an
issue. In order to solve it we can either forbit double values in a list
or just delete the duplicates. In this patch we go for the latter
solution, removing any duplicate values when creating the acquire_list.

Reviewed-by: imsnah

If a locking level wasn't specified locking used to stop. This means
that if one, for example, didn't specify anything at the LEVEL_INSTANCE
level, no locks at the LEVEL_NODE level were acquired either. With this
patch we force _LockAndExecLU to be called for all existing levels, and
break the recursion if the level doesn't exist in locking.LEVELS.

Reviewed-by: imsnah

The check for the reboot type can be done without any locks held, so
we'll move it to ExpandNames. Plus, we note in a FIXME that if the
reboot type is not full, we can probably just lock the primary node, and
leave the secondary unlocked.

Reviewed-by: imsnah

We no longer use YAML in Ganeti at all. This patch converts the QA
configuration from YAML to JSON. JSON doesn't support comments and
I had to use a hack with fields starting with '#'.

Reviewed-by: ultrotter

Reviewed-by: schreiberal

By using this Linux-specific way we don't have to care about removing the
socket file when quitting or starting (after an unclean shutdown). For a
more detailed description, see the comment in the patch.

Reviewed-by: schreiberal

This patch also sorts the list.

Reviewed-by: schreiberal

Reviewed-by: ultrotter

The cluster no longer keeps individual host's SSH key, but rather
aliases all of them to the cluster name.

Reviewed-by: ultrotter

Apparently it was forgotten when import the remote API QA tests.

Reviewed-by: schreiberal

This option will be used to add nodes to the cluster without
asking the user to confirm the key. Together with key based
authentication this can be used in the QA tests.

Reviewed-by: ultrotter

This will be used to add nodes without user interaction, specifically
in QA tests.

Reviewed-by: ultrotter

I'm going to add another option and it would make maintaining
them in constants even more complicated.

Reviewed-by: ultrotter

This patch changes SshRunner.Run to pass all arguments to
SshRunner.BuildCmd. They had the same arguments before
and should stay that way. This change makes it easier
to add new or change existing arguments.

Reviewed-by: ultrotter

Reviewed-by: ultrotter

To my knowledge they're used nowhere and it's at least slightly
confusing to people adding new QA checks.

Reviewed-by: ultrotter

It's handy to make the os scripts know which hypervisor the instance is
going to run under. In order not to change the os API we pass this
information in the environment, where the os scripts can access it if
they're hypervisor-aware.

Reviewed-by: imsnah

If the user passes an env dict to RunCmd we'll override the environment
passed to the to-be-executed command with the values in the dict. This
allows us to pass arbitrary environment values to commands we run.

Reviewed-by: imsnah 

- Remove a few experiemental code lines left as comments
- Rework first disks' boot=on addition, which was calculated twice
- Remove an empty line
- Remove reference to hvm_pae which doesn't apply to kvm

Reviewed-by: imsnah

Reviewed-by: imsnah

ht_kvm.py contains the code for ganeti to work under kvm.
This patch also modifies Makefile.am to ship that file, and
lib/hypervisor/__init__.py to import it, and add kvm to the
hypervisors map.

Reviewed-by: imsnah

Add a new hypervisor type, HT_KVM, to constants, and register it in the
HYPER_TYPES set.

Reviewed-by: imsnah

This allows to configure a different path to the kvm binary. By default
/usr/bin/kvm is used, which is the one found in debian and ubuntu.

Reviewed-by: imsnah