Users - running virtual hosts via apache
Locked
12
12
 
|
Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. For the most part the clustering support comes from the Servlet container that OFBiz runs on. This is one of the reasons that we like Tomcat as the default Servlet container, it supports session replication and there are various tools you can run in front of it (usually Apache modules like mod_jk or whatever) to do load balancing and traffic routing in the case of a server failure. So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. I guess the main message I want to communicate is that part of the point of using standards-based infrastructure is so that we have one less thing to worry about in OFBiz, but that also means that OFBiz is not a one-stop-shop for all related solutions. There is pretty good documentation and even books about Tomcat, Apache and various modules to tie them together. Those are great sources of information, and people that specialize in that will typically know much more and produce a much better setup than an OFBiz-oriented professional could typically offer. That goes for book authors and such as well. Of course, there are various people and groups that know about both and can help with this sort of thing, as has been mentioned, but if your preference is to develop internal skills to manage this there are lots of very good resources available, just not inside OFBiz. And of course, if you deploy in another container the clustering configuration is totally different... I guess that's the distinction between and fun with standards-based versus standardized. ;) -David Florin T.PATRASCU (work) wrote: > Hi Andrew, > > I was looking for some more information than was/is posted to the list > and so far I found nothing clear enough to allow one to experiment > clustering with OFBiz. Or maybe I am looking from an incorrect perspective?! > > Reading Tim's reply, I understand that "clustering" looks to be a well > guarded :) secret, so we may have to contact a commercial vendor for > consulting in case we cannot figure it out by ourselves. > > I believe a clean and a step-by-step OFBiz-clustering HOWTO or similar, > will be very needed by the OFBiz community. A document like that, having > the blessing of David and Andy, will be very important for the > community. Maybe someone here has time and knowledge to start it? > > For my team right now, we're only trying to conduct some stress tests on > OFBIz, therefore will be a little bit premature for our budget to pay > somebody big money for this but it looks like we have to. > > Thanks anyway Andrew, > -florin > > > Andrew Sykes wrote: >> Florin, >> >> Well, I don't believe anyone ever did answer that question, but... >> >> There was some discussion of this a few days ago. >> >> Using distributed-cache-clear, RMI etc... >> > > > _______________________________________________ > Users mailing list > [hidden email] > http://lists.ofbiz.org/mailman/listinfo/users _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
Re: Users - OFBiz clustering
|
|
Ok then, I will go and read the documentation you are referring too and
when we'll finalize our attempt in clustering OFBiz we will try to come
back and post here our experience in case somebody else is looking for
a shortcut.
Thank you David. -florin David E. Jones wrote: Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. For the most part the clustering support comes from the Servlet container that OFBiz runs on. This is one of the reasons that we like Tomcat as the default Servlet container, it supports session replication and there are various tools you can run in front of it (usually Apache modules like mod_jk or whatever) to do load balancing and traffic routing in the case of a server failure. So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. I guess the main message I want to communicate is that part of the point of using standards-based infrastructure is so that we have one less thing to worry about in OFBiz, but that also means that OFBiz is not a one-stop-shop for all related solutions. There is pretty good documentation and even books about Tomcat, Apache and various modules to tie them together. Those are great sources of information, and people that specialize in that will typically know much more and produce a much better setup than an OFBiz-oriented professional could typically offer. That goes for book authors and such as well. Of course, there are various people and groups that know about both and can help with this sort of thing, as has been mentioned, but if your preference is to develop internal skills to manage this there are lots of very good resources available, just not inside OFBiz. And of course, if you deploy in another container the clustering configuration is totally different... I guess that's the distinction between and fun with standards-based versus standardized. ;) -David Florin T.PATRASCU (work) wrote: _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
Have you searched through the wiki? I'm pretty sure a few people have done just what you describe and posted things there... -David Florin T.PATRASCU (work) wrote: > Ok then, I will go and read the documentation you are referring too and > when we'll finalize our attempt in clustering OFBiz we will try to come > back and post here our experience in case somebody else is looking for a > shortcut. > > Thank you David. > -florin > > > David E. Jones wrote: >> Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. >> >> For the most part the clustering support comes from the Servlet container that OFBiz runs on. This is one of the reasons that we like Tomcat as the default Servlet container, it supports session replication and there are various tools you can run in front of it (usually Apache modules like mod_jk or whatever) to do load balancing and traffic routing in the case of a server failure. >> >> So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. >> >> I guess the main message I want to communicate is that part of the point of using standards-based infrastructure is so that we have one less thing to worry about in OFBiz, but that also means that OFBiz is not a one-stop-shop for all related solutions. There is pretty good documentation and even books about Tomcat, Apache and various modules to tie them together. Those are great sources of information, and people that specialize in that will typically know much more and produce a much better setup than an OFBiz-oriented professional could typically offer. That goes for book authors and such as well. >> >> Of course, there are various people and groups that know about both and can help with this sort of thing, as has been mentioned, but if your preference is to develop internal skills to manage this there are lots of very good resources available, just not inside OFBiz. >> >> And of course, if you deploy in another container the clustering configuration is totally different... I guess that's the distinction between and fun with standards-based versus standardized. ;) >> >> -David >> >> >> Florin T.PATRASCU (work) wrote: >> >>> Hi Andrew, >>> >>> I was looking for some more information than was/is posted to the list >>> and so far I found nothing clear enough to allow one to experiment >>> clustering with OFBiz. Or maybe I am looking from an incorrect perspective?! >>> >>> Reading Tim's reply, I understand that "clustering" looks to be a well >>> guarded :) secret, so we may have to contact a commercial vendor for >>> consulting in case we cannot figure it out by ourselves. >>> >>> I believe a clean and a step-by-step OFBiz-clustering HOWTO or similar, >>> will be very needed by the OFBiz community. A document like that, having >>> the blessing of David and Andy, will be very important for the >>> community. Maybe someone here has time and knowledge to start it? >>> >>> For my team right now, we're only trying to conduct some stress tests on >>> OFBIz, therefore will be a little bit premature for our budget to pay >>> somebody big money for this but it looks like we have to. >>> >>> Thanks anyway Andrew, >>> -florin >>> >>> >>> Andrew Sykes wrote: >>> >>>> Florin, >>>> >>>> Well, I don't believe anyone ever did answer that question, but... >>>> >>>> There was some discussion of this a few days ago. >>>> >>>> Using distributed-cache-clear, RMI etc... >>>> >>>> >>> >>> _______________________________________________ >>> Users mailing list >>> [hidden email] >>> http://lists.ofbiz.org/mailman/listinfo/users >>> >> >> _______________________________________________ >> Users mailing list >> [hidden email] >> http://lists.ofbiz.org/mailman/listinfo/users >> >> >> >> >> > > ------------------------------------------------------------------------ > > > _______________________________________________ > Users mailing list > [hidden email] > http://lists.ofbiz.org/mailman/listinfo/users _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
Re: Users - OFBiz clustering
|
|
Of course, the only info I found is this:
http://ofbizwiki.go-integral.com/Search.jsp?query=clustering&ok=Search I cannot comment about it, who ever added there did it with good intentions but... I don't know about others here but as far as I am concerned I was looking to check if anybody here can share some of his/her experience regarding *clustering OFBiz* not TOMCAT. But as I've said, we are going in this direction and we'll try to bring back some feedback. I hope the new OFBiz Wiki/Confluence will be better (content wise) than the old one; having a better content will make threads like this one ... disappear ;) Thanks anyway, -florin David E. Jones wrote: > Have you searched through the wiki? I'm pretty sure a few people have done just what you describe and posted things there... > > -David > > > Florin T.PATRASCU (work) wrote: > >> Ok then, I will go and read the documentation you are referring too and >> when we'll finalize our attempt in clustering OFBiz we will try to come >> back and post here our experience in case somebody else is looking for a >> shortcut. >> >> Thank you David. >> -florin >> >> >> David E. Jones wrote: >> >>> Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. >>> >>> For the most part the clustering support comes from the Servlet container that OFBiz runs on. This is one of the reasons that we like Tomcat as the default Servlet container, it supports session replication and there are various tools you can run in front of it (usually Apache modules like mod_jk or whatever) to do load balancing and traffic routing in the case of a server failure. >>> >>> So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. >>> >>> I guess the main message I want to communicate is that part of the point of using standards-based infrastructure is so that we have one less thing to worry about in OFBiz, but that also means that OFBiz is not a one-stop-shop for all related solutions. There is pretty good documentation and even books about Tomcat, Apache and various modules to tie them together. Those are great sources of information, and people that specialize in that will typically know much more and produce a much better setup than an OFBiz-oriented professional could typically offer. That goes for book authors and such as well. >>> >>> Of course, there are various people and groups that know about both and can help with this sort of thing, as has been mentioned, but if your preference is to develop internal skills to manage this there are lots of very good resources available, just not inside OFBiz. >>> >>> And of course, if you deploy in another container the clustering configuration is totally different... I guess that's the distinction between and fun with standards-based versus standardized. ;) >>> >>> -David >>> >>> >>> Florin T.PATRASCU (work) wrote: >>> >>> >>>> Hi Andrew, >>>> >>>> I was looking for some more information than was/is posted to the list >>>> and so far I found nothing clear enough to allow one to experiment >>>> clustering with OFBiz. Or maybe I am looking from an incorrect perspective?! >>>> >>>> Reading Tim's reply, I understand that "clustering" looks to be a well >>>> guarded :) secret, so we may have to contact a commercial vendor for >>>> consulting in case we cannot figure it out by ourselves. >>>> >>>> I believe a clean and a step-by-step OFBiz-clustering HOWTO or similar, >>>> will be very needed by the OFBiz community. A document like that, having >>>> the blessing of David and Andy, will be very important for the >>>> community. Maybe someone here has time and knowledge to start it? >>>> >>>> For my team right now, we're only trying to conduct some stress tests on >>>> OFBIz, therefore will be a little bit premature for our budget to pay >>>> somebody big money for this but it looks like we have to. >>>> >>>> Thanks anyway Andrew, >>>> -florin >>>> >>>> >>>> Andrew Sykes wrote: >>>> >>>> >>>>> Florin, >>>>> >>>>> Well, I don't believe anyone ever did answer that question, but... >>>>> >>>>> There was some discussion of this a few days ago. >>>>> >>>>> Using distributed-cache-clear, RMI etc... >>>>> >>>>> >>>>> >>>> >>>> _______________________________________________ >>>> Users mailing list >>>> [hidden email] >>>> http://lists.ofbiz.org/mailman/listinfo/users >>>> >>>> >>> >>> _______________________________________________ >>> Users mailing list >>> [hidden email] >>> http://lists.ofbiz.org/mailman/listinfo/users >>> >>> >>> >>> >>> >>> >> ------------------------------------------------------------------------ >> >> >> _______________________________________________ >> Users mailing list >> [hidden email] >> http://lists.ofbiz.org/mailman/listinfo/users >> > > _______________________________________________ > Users mailing list > [hidden email] > http://lists.ofbiz.org/mailman/listinfo/users > > > > > _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
In reply to this post by David E. Jones
> Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful.
... > So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. That makes sense, and sounds good - better to use what other people have already provided. Out of curiousity, though, how do things work out with the services system? Say I schedule a service for 1AM, what happens, in very high level terms? I have a feeling that low-level the result is complex enough that I'd be a week or two trying to get the answer by code-diving. -- David N. Welton - http://www.dedasys.com/davidw/ Linux, Open Source Consulting - http://www.dedasys.com/ _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
David (Welton),
Each instance would have its own job pool, so only that instance would run the service. On Mon, 2006-06-12 at 11:26 +0200, David Welton wrote: > > Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. > > ... > > > So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. > > That makes sense, and sounds good - better to use what other people > have already provided. > > Out of curiousity, though, how do things work out with the services > system? Say I schedule a service for 1AM, what happens, in very high > level terms? I have a feeling that low-level the result is complex > enough that I'd be a week or two trying to get the answer by > code-diving. > Kind Regards Andrew Sykes <[hidden email]> Sykes Development Ltd http://www.sykesdevelopment.com _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
For high volume environments you usually have a number of servers in the same service engine pool so that they can share the load of jobs. These are often the same set of machines as the web application servers. It means that a scheduled or asynchronous persisted service could be started by one machine in the group and then picked up and run by another machine. As long as the code is synchronized across the pool it doesn't matter which machine picks it up and runs it as everything persisted is in the share database anyway. -David Andrew Sykes wrote: > David (Welton), > > Each instance would have its own job pool, so only that instance would > run the service. > > On Mon, 2006-06-12 at 11:26 +0200, David Welton wrote: >>> Most of the clustering of OFBiz actually has nothing to do with OFBiz... The exceptions are things like the distributed cache clearing feature of the Entity Engine, and depending on your setup certain Service Engine configuration changes might be helpful. >> ... >> >>> So, while there is some helpful stuff in the OFBiz wiki, the best place to research this stuff is looking at these features in Tomcat, Apache, and the different modules that support load balancing and such. >> That makes sense, and sounds good - better to use what other people >> have already provided. >> >> Out of curiousity, though, how do things work out with the services >> system? Say I schedule a service for 1AM, what happens, in very high >> level terms? I have a feeling that low-level the result is complex >> enough that I'd be a week or two trying to get the answer by >> code-diving. >> _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
> For high volume environments you usually have a number of servers in the same service engine pool so that they can share the load of jobs. These are often the same set of machines as the web application servers. It means that a scheduled or asynchronous persisted service could be started by one machine in the group and then picked up and run by another machine. As long as the code is synchronized across the pool it doesn't matter which machine picks it up and runs it as everything persisted is in the share database anyway.
Very nice! Thankyou for the elucidation. -- David N. Welton - http://www.dedasys.com/davidw/ Linux, Open Source Consulting - http://www.dedasys.com/ _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
David,
I looked at doing this, but couldn't quickly identify what stopped the job being run simultaneously by more than one machine. What should I be looking for? On Mon, 2006-06-12 at 18:57 +0200, David Welton wrote: > > For high volume environments you usually have a number of servers in the same service engine pool so that they can share the load of jobs. These are often the same set of machines as the web application servers. It means that a scheduled or asynchronous persisted service could be started by one machine in the group and then picked up and run by another machine. As long as the code is synchronized across the pool it doesn't matter which machine picks it up and runs it as everything persisted is in the share database anyway. > > Very nice! Thankyou for the elucidation. > -- Kind Regards Andrew Sykes <[hidden email]> Sykes Development Ltd http://www.sykesdevelopment.com _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
|
|
The service engine takes care of that. Each job has a status that is updated when the run is started. -David Andrew Sykes wrote: > David, > > I looked at doing this, but couldn't quickly identify what stopped the > job being run simultaneously by more than one machine. > > What should I be looking for? > > On Mon, 2006-06-12 at 18:57 +0200, David Welton wrote: >>> For high volume environments you usually have a number of servers in the same service engine pool so that they can share the load of jobs. These are often the same set of machines as the web application servers. It means that a scheduled or asynchronous persisted service could be started by one machine in the group and then picked up and run by another machine. As long as the code is synchronized across the pool it doesn't matter which machine picks it up and runs it as everything persisted is in the share database anyway. >> Very nice! Thankyou for the elucidation. >> _______________________________________________ Users mailing list [hidden email] http://lists.ofbiz.org/mailman/listinfo/users |
«
Return to OFBiz - User
|
1 view|%1 views
| Free forum by Nabble | Edit this page |