The heartbeats frequency has to be lower than the failure detection timeout. This is why we decided to calculate the former basing on a value set for the latter rather than making a user to adjust both properties properly. This is how both parameters became related some time ago :)
Honestly, I don’t think that the javadoc improvement will make things clearer for the users. Hope you will agree that people pay attention to the naming first and, only if the naming makes sense to them, learn more about the details referring to the javadoc. — Denis > On Feb 27, 2017, at 2:59 PM, Dmitriy Setrakyan <dsetrak...@apache.org> wrote: > > Hm... I don't think that heartbeat frequency has to be associated with > failure detection. What if we just update the javadoc for this parameter, > stating that it has to do with metrics update only? > > On Mon, Feb 27, 2017 at 11:44 AM, Denis Magda <dma...@apache.org> wrote: > >> Igniters, >> >> Long time ago we updated the logic in discovery SPI that issues heartbeats >> messages from one node to another. Presently, heartbeats frequency is >> calculated basing on IgniteConfiguration.failureDetectionTimeout meaning >> that TcpDiscoverySpi.heartbeatsFrequency has nothing to do with >> heartbeats frequency at all. >> >> TcpDiscoverySpi.heartbeatsFrequency defines a frequency for metrics >> message. So, I propose to rename this method in Apache Igntie 2.0 to >> something more meaningful like TcpDiscoverySpi.metricsUpdateFrequency? >> >> Do you agree? Alternatives thoughts? >> >> — >> Denis
