Provided a patch - https://github.com/apache/ignite/pull/2285, the
change is straight-forward - simply placed PluginProvider#onIgniteStart
right after components #onKernalStart callbacks are called. In basic
test suite (as I can see it includes platform and plugin tests) -
http://ci.ignite.apache.org/viewLog.html?buildId=720722&tab=buildResultsDiv&buildTypeId=IgniteTests_IgniteBasic
there seem to be no new failed platform tests, other failed tests seem
to fail in several other reviews too and be unrelated to my changes.
Would you be so kind to review my pull-request?
Best regards,
Evgeniy.
On 05.07.2017 12:35, Evgeniy Ignatiev wrote:
Would you mind if I pick the
https://issues.apache.org/jira/browse/IGNITE-5123 and place
PluginProvider#onIgniteStart after the component callbacks are called
to comply with its Javadoc?
On 04.07.2017 18:06, Evgeniy Ignatiev wrote:
As it is explicitly stated in Javadoc of PluginProvider#onIgniteStart
that all internal components should be started before this method is
called, which is not true, as internal component callbacks are called
right after the PluginProvider#onIgniteStart.
On 7/2/2017 12:58 PM, Evgeniy Ignatiev wrote:
Hello, Igniters.
Looking at the ticket
https://issues.apache.org/jira/browse/IGNITE-5123, the
IgniteKernal#start notifies plugin providers, calling
PluginProvider#onIgniteStart, before components are notified, when
caches are created in GridCacheProcessor#onIgniteStart. Does it make
sense to move plugin providers notification after components are
notified, probably right before lifecycle beans notification, so to
ensure that caches are started before PluginProvider#onIgniteStart
method is called?
