statisticsEnabled property comes from JCache, BTW.

-Val

On Sat, Sep 9, 2017 at 11:09 AM, Dmitriy Setrakyan <[email protected]>
wrote:

> On Sat, Sep 9, 2017 at 8:56 AM, Denis Magda <[email protected]> wrote:
>
> > Surprise!
> >
> > If you want to see cache events then you have to enable one more flag!
> >
> >  <property name="StatisticsEnabled" value="true"/>
>
>
> What is the overhead of this statistics collection?
>
>
> > Three flags/beans have to be in the config in total, three! Just to see
> > cache events. The API is a mess. Let’s contemplate how to fix it.
>
>
> Agree, this is horrible. We need to fix it in 2.3. Is there a ticket?
>
>
> >
> > —
> > Denis
> >
> > > On Sep 7, 2017, at 7:33 PM, Dmitriy Setrakyan <[email protected]>
> > wrote:
> > >
> > > On Thu, Sep 7, 2017 at 7:30 PM, Denis Magda <[email protected]> wrote:
> > >
> > >> My point is different. Before I had to do this only assuming that
> > “Ignite
> > >> will spend 99%” sending events:
> > >>
> > >>
> > >>        <property name="includeEventTypes">
> > >>            <list>
> > >>                <!--Task execution events-->
> > >>                <util:constant static-field="org.apache.
> > >> ignite.events.EventType.EVT_TASK_STARTED"/>
> > >>                <util:constant static-field="org.apache.
> > >> ignite.events.EventType.EVT_TASK_FINISHED"/>
> > >>                <util:constant static-field="org.apache.
> > >> ignite.events.EventType.EVT_TASK_FAILED"/>
> > >>                <util:constant static-field="org.apache.
> > >> ignite.events.EventType.EVT_TASK_TIMEDOUT"/>
> > >>            </list>
> > >>        </property>
> > >>
> > >> Now the platform forces me to do that (probably thinking that I’m
> crazy
> > if
> > >> I want to waste resources for metrics and giving one more change to
> > >> contemplate the decision):
> > >>
> > >>        <property name="eventStorageSpi">
> > >>           <bean class="org.apache.ignite.spi.eventstorage.memory.
> > >> MemoryEventStorageSpi"/>
> > >>       </property>
> > >>
> > >> Does the issue make sense to you know?
> > >>
> > >
> > > I understand now. Why did we change this behavior? Can someone comment?
> >
> >
>

Reply via email to