+1 Supun but it seems to me we need to involve the load balancer in order to achieve this. We have to sort this out for Stratos' load balancer anyway as its critical to allow one tenant to upgrade something (for possibly just a few users) instead of forcing a global update.
Sanjiva. On Sun, Dec 5, 2010 at 6:20 PM, Supun Kamburugamuva <[email protected]> wrote: > I think we need a feature from registry based repo for doing this > successfully in a real production scenario. > > In a production scenarion if there is a cluster of ESB's user don't want to > take down all the ESB's or update all the ESB's at the same time. Usually > when a change is made it is done in only one node. Then that node stays like > that for few days to make sure that the change is ok. During this time the > other nodes should act as previously. So for a production deployment we > shouldn't do automatic repository syncs for all the nodes by default. > > User should be able to select when to sync with the registry repo. We can > provide an UI for doing this. > > Here are the steps I think a user should follow to do a update in a > cluster. > > 1. First user has to take the master node off the traffic and update the > master node. > 3. After he thinks that the configuration is ok, enable the registry sync > in master node so that the registry is updated. > 2. Now when user want to do the change for a another ESB he has to take > that ESB out of traffic from load balancer and enable the registry sync. > This will update the configuration of that node. > > Like wise user can do a successful configuration update without restarting > any of the machines. > > Thanks, > Supun.. > > > On Sun, Dec 5, 2010 at 12:11 AM, Hiranya Jayathilaka <[email protected]>wrote: > >> >> >> On Sat, Dec 4, 2010 at 10:41 PM, Senaka Fernando <[email protected]> wrote: >> >>> >>> >>> On Sat, Dec 4, 2010 at 10:09 PM, Ruwan Linton <[email protected]> wrote: >>> >>>> On 12/4/10 8:54 PM, Supun Kamburugamuva wrote: >>>> >>>> >>>> >>>> On Sat, Dec 4, 2010 at 5:45 PM, Sanjiva Weerawarana >>>> <[email protected]>wrote: >>>> >>>>> On Sat, Dec 4, 2010 at 2:45 PM, Supun Kamburugamuva >>>>> <[email protected]>wrote: >>>>> >>>>>> >>>>>>> >>>>>>> The registry based repo, when enabled syncs the >>>>>> axis2 repository periodically. Basically it syncs what ever files that >>>>>> are >>>>>> in the axis2 repository of a cluster of servers that are connected to a >>>>>> single configuration registry. It stores the files in to the registry and >>>>>> then sync the file from there. Now we have moved the synapse >>>>>> configuration >>>>>> to the axis2 repository as well. So when a synapse configuration file is >>>>>> modified the changes will be synched across the cluster. Because we have >>>>>> hot >>>>>> update these changes will immediately taken in to account without >>>>>> requiring >>>>>> a restart of the read-only nodes. >>>>>> >>>>> >>>>> Are we going to allow any node's config to be edited and have that >>>>> sync across all nodes or do we still have a master node concept? >>>>> >>>> >>>> I think we should allow only the master node to be changed. There are >>>> some part of the configuration that doesn't work without this mode. For >>>> example we store things like security policy information about a service in >>>> the registry. So in-order for these things to work user has to always >>>> change >>>> the master node. >>>> >>>> Why? >>>> >>> >>> I see Master-node as a concept that we've introduced to overcome some of >>> the limitations we had in the past. If this model works, and if this is >>> expected to scale, we should not have a concept of a master-node in this >>> sense, as Ruwan points out. Somebody, needs to be the source of control, so >>> in that sense we could have a master. >>> >> >> +1 and +1 to disabling registry persistence for Synapse configuration. >> With the new model old registry persistence becomes obsolete really. >> >> Thanks, >> Hiranya >> >> >>> >>> Thanks, >>> Senaka. >>> >>> >>>> Any explanation why those changes has to be on the master node, what is >>>> special about the master node apart from the fact that it controls the >>>> configuration? >>>> >>> Ruwan >>>> >>>> >>>> Thanks, >>>> Supun.. >>>> >>>> >>>>> >>>>> I think we have some issues in Stratos because the registry repo is >>>>> synced on a timer. IIRC we discussed changing that to use registry >>>>> eventing >>>>> to notify the subscribers and they'll pull the changes in. Have we >>>>> implemented that? >>>>> >>>>> We need to avoid getting into a change loop as a result too .. but >>>>> we've dealt with that at a customer setup a while ago. >>>>> >>>>> Sanjiva. >>>>> -- >>>>> Sanjiva Weerawarana, Ph.D. >>>>> Founder, Chairman & CEO; WSO2, Inc.; http://wso2.com/ >>>>> email: [email protected]; phone: +94 11 763 9614; cell: +94 77 787 6880 >>>>> | +1 650 265 8311 >>>>> blog: http://sanjiva.weerawarana.org/ >>>>> >>>>> Lean . Enterprise . Middleware >>>>> >>>>> _______________________________________________ >>>>> Carbon-dev mailing list >>>>> [email protected] >>>>> https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev >>>>> >>>>> >>>> >>>> >>>> -- >>>> Supun Kamburugamuva >>>> Technical Lead >>>> WSO2 Inc.; http://wso2.org >>>> E-mail: [email protected]; Mobile: +94 77 431 3585 >>>> Blog: http://supunk.blogspot.com >>>> >>>> >>>> _______________________________________________ >>>> Carbon-dev mailing >>>> [email protected]https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev >>>> >>>> >>>> >>>> -- >>>> Ruwan Linton >>>> Software Architect & Product Manager, WSO2 ESB; http://wso2.org/esb >>>> >>>> WSO2 Inc.; http://wso2.com >>>> >>>> >>>> Lean . Enterprise . Middleware >>>> >>>> phone: +1 408 754 7388 ext 51789 >>>> email: [email protected]; cell: +94 77 341 3097 >>>> blog: http://blog.ruwan.org >>>> linkedin: http://www.linkedin.com/in/ruwanlinton >>>> tweet: http://twitter.com/ruwanlinton >>>> >>>> >>>> _______________________________________________ >>>> Carbon-dev mailing list >>>> [email protected] >>>> https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev >>>> >>>> >>> >>> >>> -- >>> *Senaka Fernando* >>> Associate Technical Lead & Product Manager - WSO2 G-Reg; >>> WSO2, Inc.; http://wso2.com >>> * >>> Member; Apache Software Foundation; http://apache.org >>> >>> E-mail: senaka AT wso2.com >>> **P: +1 408 754 7388; ext: 51736*; >>> *M: +94 77 322 1818 >>> Linked-In: http://www.linkedin.com/in/senakafernando >>> >>> *Lean . Enterprise . Middleware >>> >>> >>> _______________________________________________ >>> Carbon-dev mailing list >>> [email protected] >>> https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev >>> >>> >> >> >> -- >> Hiranya Jayathilaka >> Senior Software Engineer; >> WSO2 Inc.; http://wso2.org >> E-mail: [email protected]; Mobile: +94 77 633 3491 >> Blog: http://techfeast-hiranya.blogspot.com >> >> _______________________________________________ >> Carbon-dev mailing list >> [email protected] >> https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev >> >> > > > -- > Supun Kamburugamuva > Technical Lead > WSO2 Inc.; http://wso2.org > E-mail: [email protected]; Mobile: +94 77 431 3585 > Blog: http://supunk.blogspot.com > > _______________________________________________ > Carbon-dev mailing list > [email protected] > https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev > > -- Sanjiva Weerawarana, Ph.D. Founder, Chairman & CEO; WSO2, Inc.; http://wso2.com/ email: [email protected]; phone: +94 11 763 9614; cell: +94 77 787 6880 | +1 650 265 8311 blog: http://sanjiva.weerawarana.org/ Lean . Enterprise . Middleware
_______________________________________________ Carbon-dev mailing list [email protected] https://wso2.org/cgi-bin/mailman/listinfo/carbon-dev
