“_SPICE_ADDITIONAL_PARAMS” is supposed to be the full list of
SPICE-related KVM hypervisor parameters with the exception of
“HV_KVM_SPICE_BIND”. The new test checks if all parameters starting with
“HV_KVM_SPICE_*” are included. Three previously missing parameters are
added to the list.

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

This list is static at runtime and doesn't need to be recreated every
time.

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

When non-DRBD disks are used for an instance,
“lu.needed_locks[locking.LEVEL_NODE]” is set to “locking.ALL_SET” (which
is None). The assertion will then fail as None evaluates to False.

Reported by Constantinos Venetsanopoulos.

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

Doing so avoids triggering an assertion in mcpu which cross-checks the
node lock and node allocation lock acquisition mode.

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

Commit f0f8d060 (“Show old primary/secondary node on disk
replacement”) added two wrong uses of feedback_fn, which results in
log entries such like these:

      "log": [
        [
          7, [1351258326, 466214], "message", "Replacing disk(s) 0 for instance 'instance1.example.com'"
        ],
        [
          8, [1351258326, 749391], "Current primary node: %s", "node1"
        ],
        [
          9, [1351258326, 774829], "Current seconary node: %s", "node4"
        ],
        [
          10, [1351258329, 620410], "message", "STEP 1/6 Check device existence"
        ],

(the first and fourth are OK, the 2nd and 3rd are broken).

The API of jqueue._OpExecCallbacks.Feedback should probably be changed
to prevent such an occurrence in the future.

Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Michele Tartara <mtartara@google.com>

For some opcodes, the output is not "stable", and depends on the exact
input values; this makes it harder to check consistency against
Haskell code.

To compensate for this, we add a way to override the formatting of the
OP_DSC_FIELD; by default, this is always "%s", but if the
OP_DSC_FORMATTER is defined (must be a callable), it is used to format
the actual value.

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

Introduced in commit d4752785.

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

This is a "better than nothing" support, just for kvm and just joining
the machine to the opevswitch bridge with the right command.

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

This should be done because in the case of --no-name-check there
is no 'hostname.ip' attribute, causing an execution error.
'check_ip' is always set (in CheckArguments) even if --no-name-check
is passed in the command line.

Signed-off-by: Constantinos Venetsanopoulos <cven@grnet.gr>
Signed-off-by: Iustin Pop <iustin@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

The master daemon should not directly read files written by the node
daemon. This patch adds a new RPC to read the watcher pause file and
changes the master code to use it.

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

Instead of just setting the watcher pause file, which isn't replicated,
RPC is used to set it on all nodes (where possible). This means that
after an outage of the master node and a following master-failover, the
watcher will still be paused.

The RPC is attempted on all nodes. Errors for nodes marked offline are
ignored. All other errors are returned in the form of an exception
(there is no other option given the LUXI API).

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

The watcher pause file should be set/unset on all nodes at once, not
only the master node. For that a new RPC is needed.

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

This addresses issue 218. When the number of inotify watches is
exhausted, for example by being set too low from the beginning or by
other programs, waiting for a job to change would just report a lost job
(e.g. “Error checking job status: Job with id 7817 lost”).

This patch changes the job watcher to no longer catch
“errors.InotifyError” and, this is by far the larger part of this patch,
adds unittests for this situation.

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

- Add more checks, some of them are deliberately redundant
- Descriptive error messages
- Add comment describing order to “tools.ensure_dirs”
- Avoid copying a list in an assertion in “tools.ensure_dirs”

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

Doing so simplifies to code a bit and never had a practical use.

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

Change the "--disks" option validation, to just check the format
of the dict and do not check whether the keys are included in the
IDISK_PARAMS constant at OpCode level. This allows the passing of
arbitrary parameters at the CLI, which will then be logically
checked at LU level (inside CheckArguments).

Currently, the check wrt to IDISK_PARAMS already happens in all
related LUs (InstanceSetParams, InstanceCreate, InstanceRecreateDisks),
with the use of ForceDictType(), so there is no need for additional
checking code.

Signed-off-by: Constantinos Venetsanopoulos <cven@grnet.gr>
Signed-off-by: Iustin Pop <iustin@google.com>
[iustin@google.com: fixed a unittest that needed adjustement]
Reviewed-by: Iustin Pop <iustin@google.com>

It will be needed by the DRBD data collector, that will be added shortly.

Signed-off-by: Michele Tartara <mtartara@google.com>
Reviewed-by: Iustin Pop <iustin@google.com>

Commit ebd437a4 added two new entries to tools.ensure_dirs, but did so in
the wrong order. Patch forthcoming to improve the unittest's error
message.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Helga Velroyen <helgav@google.com>

If the files were created by a user different, e.g. due to a switch from
running masterd as root to running it as a dedicated user, they couldn't
be modified/removed anymore.

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

Wrap lines in a consistent manner (uid/gid on the second line) if
wrapping is necessary at all. “git diff --color-words” shows no
difference at all as only whitespace changed.

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

Until now, the flag was unset on a master failover unless the
“$localstatedir/lib/ganeti/queue/drain” file existed.

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

When used instead of a plain call to “frozenset”, this would have
avoided the issue fixed in commit e2dd6ece. The new function is located
in the “compat” module as it will be used at module load time in most
places and should therefore reside in a place with very few
dependencies.

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>

Unfortunately there was a bug in commit 518a45e5 whereby ECODE_TEMP_NORES
had the same value as ECODE_NORES, leading to failures in a Haskell
test. Of course this would also have affected other users of the constant.

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: Helga Velroyen <helgav@google.com>

Explicitely mention the fs.inotify.max_user_watches sysctl value.

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

Until now the error codes were not available from the RAPI client
module. A newly added unit test ensures all error codes are contained in
“ECODE_ALL”, as well as ensuring consistency between the RAPI client and
the authoritative “errors” module.

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

When an instance creation uses opportunistic locks, the iallocator might
not be able to find an allocation solution if not enough node locks (or
a suboptimal subset thereof) were acquired. As per the design document
<doc/design-opportunistic-locking.rst> a new error code denoting
temporary failures is added.

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

Also actually start using opportunistic locks (if requested).

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

Adds a new parameter to “OpInstanceCreate” and “OpInstanceMultiAlloc” to
use opportunistic locks.

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

Forward the node whitelist to the iallocator plugin.

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

Add verification code to mcpu to check an LU's locks. Two whitelists are
provided to exclude LUs from the two tests.

Signed-off-by: Michael Hanselmann <hansmi@google.com>
Reviewed-by: Bernardo Dal Seno <bdalseno@google.com>

Block instance allocations when all node locks will be acquired.

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

Block instance allocations when all node locks will be acquired.

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

Block instance allocations when all node locks will be acquired.

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

Avoid conflicts between instance allocations.

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

Block instance allocations when all node locks will be acquired.

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

Block instance allocations when all node locks will be acquired.

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

Block instance allocations when all node locks will be acquired.

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

If no node names are given, all node locks are acquired.

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