On Wed, Sep 24, 2014 at 12:42 AM, Rajkumar Rajaratnam <[email protected]> wrote:
> > > On Tue, Sep 23, 2014 at 11:54 PM, Akila Ravihansa Perera < > [email protected]> wrote: > >> Hi Manula, >> >> >>> CEP will receive the health stats once the agent will published them. >>> Suppose the instance didn't activate due to some configuration problem. So >>> how to handle this kind of scenario. Are we use the same time threshold >>> mechanism to handle that ? >>> >> >> That's a good point! >> >> I think we cannot use the approach that we used earlier in VM scenario. >> Now we don't have direct access to individual containers. I guess we will >> need a separate CEP execution plan for handling obsolete members. >> >> @Nirmal, Raj: Any thoughts on handling obsolete container type members? >> We cannot destroy a specific container using Kubernetes API right? >> > > As mentioned earlier also, we can't kill a specific container. We are > going to keep the faulty containers. We can have an error threshold to a > cluster. For example, say 5% of containers can be faulty in a cluster. If > we exceed this threshold, we can kill all containers, and create a new > cluster. > Looks like a good plan with this faulty containers threshold. Then we will not have our resources wasted if there is a continuous error occurring. Thanks. > > Thanks. > > > > -- > Rajkumar Rajaratnam > Software Engineer | WSO2, Inc. > Mobile +94777568639 | +94783498120 > -- -- Lahiru Sandaruwan Committer and PMC member, Apache Stratos, Senior Software Engineer, WSO2 Inc., http://wso2.com lean.enterprise.middleware email: [email protected] cell: (+94) 773 325 954 blog: http://lahiruwrites.blogspot.com/ twitter: http://twitter.com/lahirus linked-in: http://lk.linkedin.com/pub/lahiru-sandaruwan/16/153/146
