| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
| |
Also reflect the changes in the tests.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
Former is preferred, per:
<http://www.linux-ha.org/doc/dev-guides/_actions.html>
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
... in Pacemaker, unlike in RGManager.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
As this is to be applied regardless the __enforce_timeouts flag
in this particular case.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
| |
- raise a warning in ccs-revitalize filter
- do a proper job in cibcompact2cib filter (during conversion)
See <http://oss.clusterlabs.org/pipermail/users/2016-January/002176.html>.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
Fix a run-check test regression along.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
Resolves: rhbz#1300050
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
... as it is a default in ocf:heartbeat:VirtualDomain anyway.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
Resolves: <https://bugzilla.redhat.com/1272592>.
Reported-by: Miroslav Lisik <mlisik@redhat.com>
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was actually a conversion bug in that these unused (just declared)
resources were promoted into ones effectively executed under new stack.
Alternative solution would be to keep them included as it used to be
plus set "is-managed" meta-attribute for them to "false", but there is
at least one deeper underlying issue: ccs_flatten is not interested in
these unemployed items in the "resources" stanza at all (they are not
validated, nor they are paired with metadata-provided defaults, etc.).
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
| |
(RG = resource group)
Interim resolution for <https://bugzilla.redhat.com/1206640>.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
For this monitor operation, either use the interval as designated in the
original rgmanager agent (sticking with the most basic=lowest depth only),
or use the default of 60 seconds (the same is automatically injected as
a default by pcs).
Resolves: <https://bugzilla.redhat.com/1272570>.
Also adjust the (unit) tests respectively.
Reported-by: Miroslav Lisik <mlisik@redhat.com>
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
... that used to drop non-deprecated cluster properties like
stonith-enabled to be dropped.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
Explicit commands are definitely easier to grasp than, e.g.,
"--clone foo=bar" hidden at the tail of a "resource add" command.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
| |
Resolves <https://bugzilla.redhat.com/1270740>.
Also extend one of the unit tests to cover clone resource (borrowed from
the reported issue).
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
|
| |
Currently, it translates deprecated default-resource-stickiness,
is-managed-default and default-action-timeout cluster properties
(options) to resource/operation default properties respectively.
More can arrive later.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
See <https://bugzilla.redhat.com/1225946#c7>.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
E.g., do not propagate the "implicit" ones like dc-version
or cluster-infrastructure.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
| |
...for both "SOMETHING non-option [non-option...]" (tail of unknown
length) and fixed-length "firm" group cases.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
Also reflect that in the unit test.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
See <https://bugzilla.redhat.com/1281359>.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
+ point the user to the root cause. It makes no sense and could be
pretty misleading when we cannot produce templates via pcs while these
are vital for the dependent resources to work properly.
True to be told, template + resource(s) can always be expressed as
fully-fledged resource (with parameters transferred down to resource
following the same [intuitive] logic) but that would defeat why this
provision was introduced in the first place, wouldn't it?
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
| |
User deserves to be fully informed about a fatal conversion obstacle.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
| |
Reported-by: Miroslav Lisik <mlisik@redhat.com>
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Based on tri-state whitelist denoting positive propagation:
- 1:1
- 1:1 but emit a warning (the meaning was changing in time)
- rename the subsys
Opened questions:
- should CMAN also be renamed to PCMK?
Also fix affected (but not directly related) unit test.
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|
|
|
|
| |
Signed-off-by: Jan Pokorný <jpokorny@redhat.com>
|