Hi Danushka, Please find my comments inline.
On Tue, Feb 23, 2016 at 8:01 PM, Danushka Fernando <danush...@wso2.com> wrote: > > On Tue, Feb 23, 2016 at 5:48 PM, Lahiru Manohara <lahir...@wso2.com> > wrote: > >> Hi All, >> >> AppCloud we are implementing deployment clean up method for Kubernetes, >> otherwise it will be a overhead for Kubernetes to keep idle deployments all >> the time. We are designing this as Jax-RS service and schedule a cron job >> for execute a shell script which calls the Jax-RS service. The deployment >> clean up will be based on data which are currently available on AppCloud >> database and MSF4J monitoring dashboard database. We are selecting not idle >> application based on MSF4J database from REQUESTS_SUMMARY_PER_MINUTE table >> based on the following logic. >> > Can't we use a task implementation here? Why this is better? > Yes we can. We decided to keep this application separate from the AppCloud because if we implement this as another scheduler task in AppCloud, sometimes it will be a performance overhead for AppCloud. > > >> IF current time - last request time for application <= threshold value >> AND average request count > threshold value >> > Is this the same threshold value or different values? I know this should > come with billing any way is this going in the tier definition? For now > where this is defined? > Different values. Currently, we will define this threshold values in a property file. But future we need to read those values based on user selected tier. > >> Reason for selecting idle applications rather than not idle applications, >> we have a situation there are no single request came to a particular >> application and it will not record in MSF4J database. So considering that >> kind of situations we are getting not idle application set from MSF4J and >> it will reduce all application set from AppCloud database. For this >> operation we are using HashSet removeAll method. >> > I think "Reason for selecting idle applications rather than not idle > applications" should be "Reason for selecting not idle applications rather > than idle applications". Any way what do you mean by MSF4J database? Can > you explain more? > >> Appreciate your suggestions on above approach. >> > "Reason for selecting not idle applications rather than idle applications" Yes, you are correct. It is not MSF4J database, it was mentioned as "MSF4J monitoring dashboard database". The database which we keeping MSF4J application monitoring statistic. Please read this e-mail "HTTP Monitoring Dashboard for Micro Services in AppCloud". > >> -- >> Best regards, >> >> *Lahiru Manohara* >> *Software Engineer* >> Mobile: +94716561576 >> WSO2 Inc. | http://wso2.com >> lean.enterprise.middleware >> >> _______________________________________________ >> Architecture mailing list >> Architecture@wso2.org >> https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture >> >> > Thanks & Regards > Danushka Fernando > Senior Software Engineer > WSO2 inc. http://wso2.com/ > Mobile : +94716332729 > > _______________________________________________ > Architecture mailing list > Architecture@wso2.org > https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture > > -- Best regards, *Lahiru Manohara* *Software Engineer* Mobile: +94716561576 WSO2 Inc. | http://wso2.com lean.enterprise.middleware
_______________________________________________ Architecture mailing list Architecture@wso2.org https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture