On Thu, Oct 8, 2015 at 11:31 AM, Vladimir Ozerov <[email protected]> wrote:
> Why are we sticked to version? If both JVM has the same major version, but > different vendors, it might be even more important concern, than different > major versions of the same vendor. > I don't think even this should matter. Java has a well defined Class format, so different vendors should not have any affect on this. However, I do agree, that we should test not only different versions, but different JDK vendors as well, e.g. Oracle JDK, Open JDK, and IBM JDK. > > On Thu, Oct 8, 2015 at 8:48 PM, Dmitriy Setrakyan <[email protected]> > wrote: > > > On Thu, Oct 8, 2015 at 9:52 AM, Andrey Gura <[email protected]> wrote: > > > > > Igniters, > > > > > > There are cases when Ignite cluster nodes work on different > environments > > > and JDKs (versions and/or vendors). GridDiscoveryManager class contains > > > check that all nodes in topology ran under JDKs with the same major > Java > > > version and throws exception if check failed. I want to replace this on > > > warning message. So I have two questions: > > > > > > 1. Is there any objections about it? > > > > > > > I agree. Ignite should allow nodes with different JDKs join the cluster. > > > > > > > 2. What message will be more informative than current one ("Local > node's > > > java major version is different from remote node's one > > > [locJvmMajVer=<locJvmMajVer>, rmtJvmMajVer=<rmtJvmMajVer>]")? > > > > > > > How about: > > --- > > Local java version is different from remote [loc=<locJvmMajVer>, > > rmt=<rmtJvmMajVer>]" > > --- > > > > > > > > > > -- > > > Andrey Gura > > > GridGain Systems, Inc. > > > www.gridgain.com > > > > > >
