Hello,
I'd like to discuss IGNITE-1071 IgniteCache.metrics() method returns local metrics <https://issues.apache.org/jira/browse/IGNITE-1071> I agree to make IgniteCache.metrics() global but want to keep MBeans local. According to specification: An MBean can represent a device, an application, or any resource that needs to be managed. So, I think it's a good point that MBean represents node instead of cluster. I case we will make MBeans global (representing whole cluster) we have no chances to see metrics of specific node which can be interesting for some reasons. For example - per node cachePuts metric can show is there any problems with affinity function or hashCode collisions. Val, Semen, could you please explain reasons of MBeans refactoring? We will loose api to gain per node metrics in this case. Thoughts? |
Could we have local and global MBeans?
Or local and global metrics inside same MBean? On Thu, Mar 24, 2016 at 6:26 PM, Anton Vinogradov <[hidden email]> wrote: > Hello, > > I'd like to discuss > IGNITE-1071 IgniteCache.metrics() method returns local metrics > <https://issues.apache.org/jira/browse/IGNITE-1071> > > I agree to make IgniteCache.metrics() global but want to keep MBeans local. > > According to specification: > An MBean can represent a device, an application, or any resource that needs > to be managed. > So, I think it's a good point that MBean represents node instead of > cluster. > > I case we will make MBeans global (representing whole cluster) we have no > chances to see metrics of specific node which can be interesting for some > reasons. > For example - per node cachePuts metric can show is there any problems with > affinity function or hashCode collisions. > > Val, Semen, could you please explain reasons of MBeans refactoring? > We will loose api to gain per node metrics in this case. > > Thoughts? > -- Alexey Kuznetsov GridGain Systems www.gridgain.com |
+ 1 to Anton and Alexey suggestions. Having both local and global MBeans
should do the trick. On Thu, Mar 24, 2016 at 2:47 PM, Alexey Kuznetsov <[hidden email]> wrote: > Could we have local and global MBeans? > Or local and global metrics inside same MBean? > > On Thu, Mar 24, 2016 at 6:26 PM, Anton Vinogradov < > [hidden email]> > wrote: > > > Hello, > > > > I'd like to discuss > > IGNITE-1071 IgniteCache.metrics() method returns local metrics > > <https://issues.apache.org/jira/browse/IGNITE-1071> > > > > I agree to make IgniteCache.metrics() global but want to keep MBeans > local. > > > > According to specification: > > An MBean can represent a device, an application, or any resource that > needs > > to be managed. > > So, I think it's a good point that MBean represents node instead of > > cluster. > > > > I case we will make MBeans global (representing whole cluster) we have no > > chances to see metrics of specific node which can be interesting for some > > reasons. > > For example - per node cachePuts metric can show is there any problems > with > > affinity function or hashCode collisions. > > > > Val, Semen, could you please explain reasons of MBeans refactoring? > > We will loose api to gain per node metrics in this case. > > > > Thoughts? > > > > > > -- > Alexey Kuznetsov > GridGain Systems > www.gridgain.com > |
Do we need local metrics at IgniteCache?
On Thu, Mar 24, 2016 at 3:02 PM, Vladimir Ozerov <[hidden email]> wrote: > + 1 to Anton and Alexey suggestions. Having both local and global MBeans > should do the trick. > > On Thu, Mar 24, 2016 at 2:47 PM, Alexey Kuznetsov <[hidden email] > > > wrote: > > > Could we have local and global MBeans? > > Or local and global metrics inside same MBean? > > > > On Thu, Mar 24, 2016 at 6:26 PM, Anton Vinogradov < > > [hidden email]> > > wrote: > > > > > Hello, > > > > > > I'd like to discuss > > > IGNITE-1071 IgniteCache.metrics() method returns local metrics > > > <https://issues.apache.org/jira/browse/IGNITE-1071> > > > > > > I agree to make IgniteCache.metrics() global but want to keep MBeans > > local. > > > > > > According to specification: > > > An MBean can represent a device, an application, or any resource that > > needs > > > to be managed. > > > So, I think it's a good point that MBean represents node instead of > > > cluster. > > > > > > I case we will make MBeans global (representing whole cluster) we have > no > > > chances to see metrics of specific node which can be interesting for > some > > > reasons. > > > For example - per node cachePuts metric can show is there any problems > > with > > > affinity function or hashCode collisions. > > > > > > Val, Semen, could you please explain reasons of MBeans refactoring? > > > We will loose api to gain per node metrics in this case. > > > > > > Thoughts? > > > > > > > > > > > -- > > Alexey Kuznetsov > > GridGain Systems > > www.gridgain.com > > > |
On Thu, Mar 24, 2016 at 5:35 AM, Anton Vinogradov <[hidden email]>
wrote: > Do we need local metrics at IgniteCache? > Yes, I don’t see any reason why not. Let’s just follow the naming pattern and prefix all local methods with the word “local”. > > On Thu, Mar 24, 2016 at 3:02 PM, Vladimir Ozerov <[hidden email]> > wrote: > > > + 1 to Anton and Alexey suggestions. Having both local and global MBeans > > should do the trick. > > > > On Thu, Mar 24, 2016 at 2:47 PM, Alexey Kuznetsov < > [hidden email] > > > > > wrote: > > > > > Could we have local and global MBeans? > > > Or local and global metrics inside same MBean? > > > > > > On Thu, Mar 24, 2016 at 6:26 PM, Anton Vinogradov < > > > [hidden email]> > > > wrote: > > > > > > > Hello, > > > > > > > > I'd like to discuss > > > > IGNITE-1071 IgniteCache.metrics() method returns local metrics > > > > <https://issues.apache.org/jira/browse/IGNITE-1071> > > > > > > > > I agree to make IgniteCache.metrics() global but want to keep MBeans > > > local. > > > > > > > > According to specification: > > > > An MBean can represent a device, an application, or any resource that > > > needs > > > > to be managed. > > > > So, I think it's a good point that MBean represents node instead of > > > > cluster. > > > > > > > > I case we will make MBeans global (representing whole cluster) we > have > > no > > > > chances to see metrics of specific node which can be interesting for > > some > > > > reasons. > > > > For example - per node cachePuts metric can show is there any > problems > > > with > > > > affinity function or hashCode collisions. > > > > > > > > Val, Semen, could you please explain reasons of MBeans refactoring? > > > > We will loose api to gain per node metrics in this case. > > > > > > > > Thoughts? > > > > > > > > > > > > > > > > -- > > > Alexey Kuznetsov > > > GridGain Systems > > > www.gridgain.com > > > > > > |
Free forum by Nabble | Edit this page |