> > >> >> If we patch a webapp, then we will do it only on master node and the >> patching process is for that. Synchronizing workers with master is a >> different task which belongs to depsync. Why do we need to mix up the >> things? May be I am missing something? >> > > No, you can't have patching only at master node and let dep-synch to take > care the rest of the patching. This will leave the system in a stage where > the slave nodes depends on dep-sych for the patching to work. If we > introduce patching process, it should be consistence across all the nodes > (servers). This is how we patch OSGi bundles. > >> >>>>>>>>>>>>> This might be a different conversation, but just to share my idea....
Why dep-synch doesn't sync the patches is because the way we have limited our own depsync model. i.e. We have used depsync only for synching so called artifacts. But assume, if we had a kernal with just an update manager like the ubuntu kernal. When we want to push something, we send an update notification and then server will update itself followed by a restart if it needed. When we want to do something to a cluster, what we do is just sending the message describing what we want to do. i.e. fetching *.jars, executing those jars(in case of a local db migration etc.) and restarting the server, deleting unwanted stuff etc. Even, we can remotely convert the whole ESB cluster to a different cluster. This would be a dumb idea, but that's how I see it personally. /Ruchira -- *Ruchira Wageesha**Associate Technical Lead* *WSO2 Inc. - lean . enterprise . middleware | wso2.com <http://wso2.com>* *email: [email protected] <[email protected]>, blog: ruchirawageesha.blogspot.com <http://ruchirawageesha.blogspot.com>, mobile: +94 77 5493444*
_______________________________________________ Architecture mailing list [email protected] https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture
