Hey folks,
It was very painful and slow, but I finally got a working almost full-rewrite of the CatalinaContainer. I need help with testing, reviews, and code improvements. Details are found in [1]. There are no functional changes, just a rewrite. I appreciate all the help and feedback. [1] https://issues.apache.org/jira/browse/OFBIZ-9392 Cheers, Taher Alkhateeb |
Hi Everyone,
In reference to [1] and the refactoring work, and after multiple different tests, I came to realize that Tomcat clustering implementation is broken (both before and after the refactoring). The reasons are complex but they essentially have to do with setting up the right parameters for the cluster. So I suspect no one is using this feature? It is therefore, my recommendation to remove clustering from CatalinaContainer which would substantially reduce complexity of both the component configuration file and the accompanying source code which would shave off at least 120 lines of code. WDYT [1] https://issues.apache.org/jira/browse/OFBIZ-9392 On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb <[hidden email]> wrote: > Hey folks, > > It was very painful and slow, but I finally got a working almost > full-rewrite of the CatalinaContainer. I need help with testing, reviews, > and code improvements. > > Details are found in [1]. There are no functional changes, just a rewrite. > I appreciate all the help and feedback. > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > Cheers, > > Taher Alkhateeb > > > |
Hi Taher,
clustering support is an important feature for enterprise installations and it worked in the past. It should be fixed if it is broken. What makes you think that clustering is broken? How did you test it? What is not working exactly? I recently updated the Tomcat version, could this be a source of the problem? Maybe there were changes in the configuration I have missed, although I did not find anything in the release notes. Regards, Michael Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: > Hi Everyone, > > In reference to [1] and the refactoring work, and after multiple different > tests, I came to realize that Tomcat clustering implementation is broken > (both before and after the refactoring). The reasons are complex but they > essentially have to do with setting up the right parameters for the > cluster. So I suspect no one is using this feature? > > It is therefore, my recommendation to remove clustering from > CatalinaContainer which would substantially reduce complexity of both the > component configuration file and the accompanying source code which would > shave off at least 120 lines of code. > > WDYT > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb <[hidden email]> > wrote: > >> Hey folks, >> >> It was very painful and slow, but I finally got a working almost >> full-rewrite of the CatalinaContainer. I need help with testing, reviews, >> and code improvements. >> >> Details are found in [1]. There are no functional changes, just a rewrite. >> I appreciate all the help and feedback. >> >> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >> >> Cheers, >> >> Taher Alkhateeb >> >> >> smime.p7s (5K) Download Attachment |
Hi Michael,
Hmmm, I didn't know that it is actively used. Essentially, you get exceptions / crashes when you uncomment the clustering section in catalina's ofbiz-component.xml. Also, I could be wrong, but I thought instead of using Tomcat's clustering, you instead cluster your infrastructure itself behind load balancers and mirroring solutions and whatnot. Anyway, if I'm wrong, then the second option is to try and figure out what's wrong with clustering and how to fix it. It's a big section of CatalinaContainer and it's currently crashing because of a missing context and a manager and a few other things. Cheers, Taher Alkhateeb On 13/06/17 01:03, Michael Brohl wrote: > Hi Taher, > > clustering support is an important feature for enterprise > installations and it worked in the past. It should be fixed if it is > broken. > > What makes you think that clustering is broken? How did you test it? > What is not working exactly? > > I recently updated the Tomcat version, could this be a source of the > problem? Maybe there were changes in the configuration I have missed, > although I did not find anything in the release notes. > > Regards, > > Michael > > > Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >> Hi Everyone, >> >> In reference to [1] and the refactoring work, and after multiple >> different >> tests, I came to realize that Tomcat clustering implementation is broken >> (both before and after the refactoring). The reasons are complex but >> they >> essentially have to do with setting up the right parameters for the >> cluster. So I suspect no one is using this feature? >> >> It is therefore, my recommendation to remove clustering from >> CatalinaContainer which would substantially reduce complexity of both >> the >> component configuration file and the accompanying source code which >> would >> shave off at least 120 lines of code. >> >> WDYT >> >> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >> >> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb >> <[hidden email]> >> wrote: >> >>> Hey folks, >>> >>> It was very painful and slow, but I finally got a working almost >>> full-rewrite of the CatalinaContainer. I need help with testing, >>> reviews, >>> and code improvements. >>> >>> Details are found in [1]. There are no functional changes, just a >>> rewrite. >>> I appreciate all the help and feedback. >>> >>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>> >>> Cheers, >>> >>> Taher Alkhateeb >>> >>> >>> > > |
Quick checked the cluster config and see that the cluster configuration
depends on org.apache.catalina.ha.session.DeltaManager, whose package is configured in Gradle as a compile dependency (org.apache.tomcat:tomcat-catalina-ha:8.5.15). Do you have this package/jar available at runtime? I cannot find it anywhere on my machine after compile. Michael Am 13.06.17 um 00:11 schrieb Taher Alkhateeb: > Hi Michael, > > Hmmm, I didn't know that it is actively used. Essentially, you get > exceptions / crashes when you uncomment the clustering section in > catalina's ofbiz-component.xml. > > Also, I could be wrong, but I thought instead of using Tomcat's > clustering, you instead cluster your infrastructure itself behind load > balancers and mirroring solutions and whatnot. > > Anyway, if I'm wrong, then the second option is to try and figure out > what's wrong with clustering and how to fix it. It's a big section of > CatalinaContainer and it's currently crashing because of a missing > context and a manager and a few other things. > > Cheers, > > Taher Alkhateeb > > > On 13/06/17 01:03, Michael Brohl wrote: >> Hi Taher, >> >> clustering support is an important feature for enterprise >> installations and it worked in the past. It should be fixed if it is >> broken. >> >> What makes you think that clustering is broken? How did you test it? >> What is not working exactly? >> >> I recently updated the Tomcat version, could this be a source of the >> problem? Maybe there were changes in the configuration I have missed, >> although I did not find anything in the release notes. >> >> Regards, >> >> Michael >> >> >> Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >>> Hi Everyone, >>> >>> In reference to [1] and the refactoring work, and after multiple >>> different >>> tests, I came to realize that Tomcat clustering implementation is >>> broken >>> (both before and after the refactoring). The reasons are complex but >>> they >>> essentially have to do with setting up the right parameters for the >>> cluster. So I suspect no one is using this feature? >>> >>> It is therefore, my recommendation to remove clustering from >>> CatalinaContainer which would substantially reduce complexity of >>> both the >>> component configuration file and the accompanying source code which >>> would >>> shave off at least 120 lines of code. >>> >>> WDYT >>> >>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>> >>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb >>> <[hidden email]> >>> wrote: >>> >>>> Hey folks, >>>> >>>> It was very painful and slow, but I finally got a working almost >>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>> reviews, >>>> and code improvements. >>>> >>>> Details are found in [1]. There are no functional changes, just a >>>> rewrite. >>>> I appreciate all the help and feedback. >>>> >>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> Cheers, >>>> >>>> Taher Alkhateeb >>>> >>>> >>>> >> >> > smime.p7s (5K) Download Attachment |
A compile dependency is also a runtime dependency in gradle (as opposed to
compileOnly). So if it is defined in gradle then it is available in runtime. On Jun 13, 2017 1:57 AM, "Michael Brohl" <[hidden email]> wrote: Quick checked the cluster config and see that the cluster configuration depends on org.apache.catalina.ha.session.DeltaManager, whose package is configured in Gradle as a compile dependency (org.apache.tomcat:tomcat-cata lina-ha:8.5.15). Do you have this package/jar available at runtime? I cannot find it anywhere on my machine after compile. Michael Am 13.06.17 um 00:11 schrieb Taher Alkhateeb: Hi Michael, > > Hmmm, I didn't know that it is actively used. Essentially, you get > exceptions / crashes when you uncomment the clustering section in > catalina's ofbiz-component.xml. > > Also, I could be wrong, but I thought instead of using Tomcat's > clustering, you instead cluster your infrastructure itself behind load > balancers and mirroring solutions and whatnot. > > Anyway, if I'm wrong, then the second option is to try and figure out > what's wrong with clustering and how to fix it. It's a big section of > CatalinaContainer and it's currently crashing because of a missing context > and a manager and a few other things. > > Cheers, > > Taher Alkhateeb > > > On 13/06/17 01:03, Michael Brohl wrote: > >> Hi Taher, >> >> clustering support is an important feature for enterprise installations >> and it worked in the past. It should be fixed if it is broken. >> >> What makes you think that clustering is broken? How did you test it? What >> is not working exactly? >> >> I recently updated the Tomcat version, could this be a source of the >> problem? Maybe there were changes in the configuration I have missed, >> although I did not find anything in the release notes. >> >> Regards, >> >> Michael >> >> >> Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >> >>> Hi Everyone, >>> >>> In reference to [1] and the refactoring work, and after multiple >>> different >>> tests, I came to realize that Tomcat clustering implementation is broken >>> (both before and after the refactoring). The reasons are complex but they >>> essentially have to do with setting up the right parameters for the >>> cluster. So I suspect no one is using this feature? >>> >>> It is therefore, my recommendation to remove clustering from >>> CatalinaContainer which would substantially reduce complexity of both the >>> component configuration file and the accompanying source code which would >>> shave off at least 120 lines of code. >>> >>> WDYT >>> >>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>> >>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>> [hidden email]> >>> wrote: >>> >>> Hey folks, >>>> >>>> It was very painful and slow, but I finally got a working almost >>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>> reviews, >>>> and code improvements. >>>> >>>> Details are found in [1]. There are no functional changes, just a >>>> rewrite. >>>> I appreciate all the help and feedback. >>>> >>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> Cheers, >>>> >>>> Taher Alkhateeb >>>> >>>> >>>> >>>> >> >> > |
Where would you expect the jar file to be stored?
I'll try to track down the errors. I was able to reduce the error messages by adding -Djava.net.preferIPv4Stack=true to the jvm arguments in build.gradle def jvmArguments = ['-Xms128M', '-Xmx1024M', "-Djava.net.preferIPv4Stack=true"] (see https://stackoverflow.com/questions/6862364/slp-java-function-causing-socketexception-due-to-ip-multicast-if). I know get one more exception org.apache.catalina.tribes.ChannelException: java.net.SocketException: Can't assign requested address; No faulty members identified. After a brief internet search, this could be a problem with my MacOS, I'll try this tomorrow. It would be great if others could check if OFBiz runs on their machine with configured cluster support. Regards, Michael Am 13.06.17 um 01:01 schrieb Taher Alkhateeb: > A compile dependency is also a runtime dependency in gradle (as opposed to > compileOnly). > > So if it is defined in gradle then it is available in runtime. > > On Jun 13, 2017 1:57 AM, "Michael Brohl" <[hidden email]> wrote: > > Quick checked the cluster config and see that the cluster configuration > depends on org.apache.catalina.ha.session.DeltaManager, whose package is > configured in Gradle as a compile dependency (org.apache.tomcat:tomcat-cata > lina-ha:8.5.15). > > Do you have this package/jar available at runtime? > > I cannot find it anywhere on my machine after compile. > > Michael > > > Am 13.06.17 um 00:11 schrieb Taher Alkhateeb: > > Hi Michael, >> Hmmm, I didn't know that it is actively used. Essentially, you get >> exceptions / crashes when you uncomment the clustering section in >> catalina's ofbiz-component.xml. >> >> Also, I could be wrong, but I thought instead of using Tomcat's >> clustering, you instead cluster your infrastructure itself behind load >> balancers and mirroring solutions and whatnot. >> >> Anyway, if I'm wrong, then the second option is to try and figure out >> what's wrong with clustering and how to fix it. It's a big section of >> CatalinaContainer and it's currently crashing because of a missing context >> and a manager and a few other things. >> >> Cheers, >> >> Taher Alkhateeb >> >> >> On 13/06/17 01:03, Michael Brohl wrote: >> >>> Hi Taher, >>> >>> clustering support is an important feature for enterprise installations >>> and it worked in the past. It should be fixed if it is broken. >>> >>> What makes you think that clustering is broken? How did you test it? What >>> is not working exactly? >>> >>> I recently updated the Tomcat version, could this be a source of the >>> problem? Maybe there were changes in the configuration I have missed, >>> although I did not find anything in the release notes. >>> >>> Regards, >>> >>> Michael >>> >>> >>> Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >>> >>>> Hi Everyone, >>>> >>>> In reference to [1] and the refactoring work, and after multiple >>>> different >>>> tests, I came to realize that Tomcat clustering implementation is broken >>>> (both before and after the refactoring). The reasons are complex but they >>>> essentially have to do with setting up the right parameters for the >>>> cluster. So I suspect no one is using this feature? >>>> >>>> It is therefore, my recommendation to remove clustering from >>>> CatalinaContainer which would substantially reduce complexity of both the >>>> component configuration file and the accompanying source code which would >>>> shave off at least 120 lines of code. >>>> >>>> WDYT >>>> >>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>> [hidden email]> >>>> wrote: >>>> >>>> Hey folks, >>>>> It was very painful and slow, but I finally got a working almost >>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>> reviews, >>>>> and code improvements. >>>>> >>>>> Details are found in [1]. There are no functional changes, just a >>>>> rewrite. >>>>> I appreciate all the help and feedback. >>>>> >>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> Cheers, >>>>> >>>>> Taher Alkhateeb >>>>> >>>>> >>>>> >>>>> >>> smime.p7s (5K) Download Attachment |
All dependencies are stored in the gradle cache and referenced in the
generated ofbiz.jar. So I guess it's somewhere in the cache. Can't you find it there? On Jun 13, 2017 2:49 AM, "Michael Brohl" <[hidden email]> wrote: Where would you expect the jar file to be stored? I'll try to track down the errors. I was able to reduce the error messages by adding -Djava.net.preferIPv4Stack=true to the jvm arguments in build.gradle def jvmArguments = ['-Xms128M', '-Xmx1024M', "-Djava.net.preferIPv4Stack=tr ue"] (see https://stackoverflow.com/questions/6862364/slp-java-functio n-causing-socketexception-due-to-ip-multicast-if). I know get one more exception org.apache.catalina.tribes.ChannelException: java.net.SocketException: Can't assign requested address; No faulty members identified. After a brief internet search, this could be a problem with my MacOS, I'll try this tomorrow. It would be great if others could check if OFBiz runs on their machine with configured cluster support. Regards, Michael Am 13.06.17 um 01:01 schrieb Taher Alkhateeb: A compile dependency is also a runtime dependency in gradle (as opposed to > compileOnly). > > So if it is defined in gradle then it is available in runtime. > > On Jun 13, 2017 1:57 AM, "Michael Brohl" <[hidden email]> wrote: > > Quick checked the cluster config and see that the cluster configuration > depends on org.apache.catalina.ha.session.DeltaManager, whose package is > configured in Gradle as a compile dependency (org.apache.tomcat:tomcat-cata > lina-ha:8.5.15). > > Do you have this package/jar available at runtime? > > I cannot find it anywhere on my machine after compile. > > Michael > > > Am 13.06.17 um 00:11 schrieb Taher Alkhateeb: > > Hi Michael, > >> Hmmm, I didn't know that it is actively used. Essentially, you get >> exceptions / crashes when you uncomment the clustering section in >> catalina's ofbiz-component.xml. >> >> Also, I could be wrong, but I thought instead of using Tomcat's >> clustering, you instead cluster your infrastructure itself behind load >> balancers and mirroring solutions and whatnot. >> >> Anyway, if I'm wrong, then the second option is to try and figure out >> what's wrong with clustering and how to fix it. It's a big section of >> CatalinaContainer and it's currently crashing because of a missing context >> and a manager and a few other things. >> >> Cheers, >> >> Taher Alkhateeb >> >> >> On 13/06/17 01:03, Michael Brohl wrote: >> >> Hi Taher, >>> >>> clustering support is an important feature for enterprise installations >>> and it worked in the past. It should be fixed if it is broken. >>> >>> What makes you think that clustering is broken? How did you test it? What >>> is not working exactly? >>> >>> I recently updated the Tomcat version, could this be a source of the >>> problem? Maybe there were changes in the configuration I have missed, >>> although I did not find anything in the release notes. >>> >>> Regards, >>> >>> Michael >>> >>> >>> Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >>> >>> Hi Everyone, >>>> >>>> In reference to [1] and the refactoring work, and after multiple >>>> different >>>> tests, I came to realize that Tomcat clustering implementation is broken >>>> (both before and after the refactoring). The reasons are complex but >>>> they >>>> essentially have to do with setting up the right parameters for the >>>> cluster. So I suspect no one is using this feature? >>>> >>>> It is therefore, my recommendation to remove clustering from >>>> CatalinaContainer which would substantially reduce complexity of both >>>> the >>>> component configuration file and the accompanying source code which >>>> would >>>> shave off at least 120 lines of code. >>>> >>>> WDYT >>>> >>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>> [hidden email]> >>>> wrote: >>>> >>>> Hey folks, >>>> >>>>> It was very painful and slow, but I finally got a working almost >>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>> reviews, >>>>> and code improvements. >>>>> >>>>> Details are found in [1]. There are no functional changes, just a >>>>> rewrite. >>>>> I appreciate all the help and feedback. >>>>> >>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> Cheers, >>>>> >>>>> Taher Alkhateeb >>>>> >>>>> >>>>> >>>>> >>>>> >>> |
No, I cannot find it. But there's no error in reference to a missing
manager class so I guess it's not a problem. If I can solve the address problem, I'll see further. Thanks and good night, Michael Am 13.06.17 um 01:51 schrieb Taher Alkhateeb: > All dependencies are stored in the gradle cache and referenced in the > generated ofbiz.jar. So I guess it's somewhere in the cache. Can't you find > it there? > > On Jun 13, 2017 2:49 AM, "Michael Brohl" <[hidden email]> wrote: > > Where would you expect the jar file to be stored? > > I'll try to track down the errors. I was able to reduce the error messages > by adding -Djava.net.preferIPv4Stack=true to the jvm arguments in > build.gradle > > def jvmArguments = ['-Xms128M', '-Xmx1024M', "-Djava.net.preferIPv4Stack=tr > ue"] > > (see https://stackoverflow.com/questions/6862364/slp-java-functio > n-causing-socketexception-due-to-ip-multicast-if). > > I know get one more exception > > org.apache.catalina.tribes.ChannelException: java.net.SocketException: > Can't assign requested address; No faulty members identified. > > After a brief internet search, this could be a problem with my MacOS, I'll > try this tomorrow. > > > It would be great if others could check if OFBiz runs on their machine with > configured cluster support. > > Regards, > > Michael > > > Am 13.06.17 um 01:01 schrieb Taher Alkhateeb: > > A compile dependency is also a runtime dependency in gradle (as opposed to >> compileOnly). >> >> So if it is defined in gradle then it is available in runtime. >> >> On Jun 13, 2017 1:57 AM, "Michael Brohl" <[hidden email]> wrote: >> >> Quick checked the cluster config and see that the cluster configuration >> depends on org.apache.catalina.ha.session.DeltaManager, whose package is >> configured in Gradle as a compile dependency (org.apache.tomcat:tomcat-cata >> lina-ha:8.5.15). >> >> Do you have this package/jar available at runtime? >> >> I cannot find it anywhere on my machine after compile. >> >> Michael >> >> >> Am 13.06.17 um 00:11 schrieb Taher Alkhateeb: >> >> Hi Michael, >> >>> Hmmm, I didn't know that it is actively used. Essentially, you get >>> exceptions / crashes when you uncomment the clustering section in >>> catalina's ofbiz-component.xml. >>> >>> Also, I could be wrong, but I thought instead of using Tomcat's >>> clustering, you instead cluster your infrastructure itself behind load >>> balancers and mirroring solutions and whatnot. >>> >>> Anyway, if I'm wrong, then the second option is to try and figure out >>> what's wrong with clustering and how to fix it. It's a big section of >>> CatalinaContainer and it's currently crashing because of a missing context >>> and a manager and a few other things. >>> >>> Cheers, >>> >>> Taher Alkhateeb >>> >>> >>> On 13/06/17 01:03, Michael Brohl wrote: >>> >>> Hi Taher, >>>> clustering support is an important feature for enterprise installations >>>> and it worked in the past. It should be fixed if it is broken. >>>> >>>> What makes you think that clustering is broken? How did you test it? What >>>> is not working exactly? >>>> >>>> I recently updated the Tomcat version, could this be a source of the >>>> problem? Maybe there were changes in the configuration I have missed, >>>> although I did not find anything in the release notes. >>>> >>>> Regards, >>>> >>>> Michael >>>> >>>> >>>> Am 12.06.17 um 22:57 schrieb Taher Alkhateeb: >>>> >>>> Hi Everyone, >>>>> In reference to [1] and the refactoring work, and after multiple >>>>> different >>>>> tests, I came to realize that Tomcat clustering implementation is broken >>>>> (both before and after the refactoring). The reasons are complex but >>>>> they >>>>> essentially have to do with setting up the right parameters for the >>>>> cluster. So I suspect no one is using this feature? >>>>> >>>>> It is therefore, my recommendation to remove clustering from >>>>> CatalinaContainer which would substantially reduce complexity of both >>>>> the >>>>> component configuration file and the accompanying source code which >>>>> would >>>>> shave off at least 120 lines of code. >>>>> >>>>> WDYT >>>>> >>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>>> [hidden email]> >>>>> wrote: >>>>> >>>>> Hey folks, >>>>> >>>>>> It was very painful and slow, but I finally got a working almost >>>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>>> reviews, >>>>>> and code improvements. >>>>>> >>>>>> Details are found in [1]. There are no functional changes, just a >>>>>> rewrite. >>>>>> I appreciate all the help and feedback. >>>>>> >>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>> >>>>>> Cheers, >>>>>> >>>>>> Taher Alkhateeb >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> smime.p7s (5K) Download Attachment |
In reply to this post by taher
The project I'm currently on is actively using clustering (session
replication) and I know for certain this isn't the only project nor are we the only vendor to use it. I'm using an older version of OFBiz though so I can't confirm if the latest codebase has it working. It can be tricky to set up and does have a tendency to crash ofbiz during startup if you don't have it configured correctly. Definitely a -1 for removal. Regards Scott On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> wrote: > Hi Everyone, > > In reference to [1] and the refactoring work, and after multiple different > tests, I came to realize that Tomcat clustering implementation is broken > (both before and after the refactoring). The reasons are complex but they > essentially have to do with setting up the right parameters for the > cluster. So I suspect no one is using this feature? > > It is therefore, my recommendation to remove clustering from > CatalinaContainer which would substantially reduce complexity of both the > component configuration file and the accompanying source code which would > shave off at least 120 lines of code. > > WDYT > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < > [hidden email]> > wrote: > > > Hey folks, > > > > It was very painful and slow, but I finally got a working almost > > full-rewrite of the CatalinaContainer. I need help with testing, reviews, > > and code improvements. > > > > Details are found in [1]. There are no functional changes, just a > rewrite. > > I appreciate all the help and feedback. > > > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > > > Cheers, > > > > Taher Alkhateeb > > > > > > > |
So the good thing is that the refactoring exercise surfaced this issue.
Based on feedback from Scott and Michael I suspect that the issue might be perhaps minor such as a missing configuration or dependency that's causing all these errors. I'll try to dig out the root cause and your help would be great to try and zoom in on the problem preferably on the refactored code, but even finding the root cause on existing code would greatly help. Cheers, Taher On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> wrote: > The project I'm currently on is actively using clustering (session > replication) and I know for certain this isn't the only project nor are we > the only vendor to use it. I'm using an older version of OFBiz though so I > can't confirm if the latest codebase has it working. It can be tricky to > set up and does have a tendency to crash ofbiz during startup if you don't > have it configured correctly. > > Definitely a -1 for removal. > > Regards > Scott > > On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> > wrote: > > > Hi Everyone, > > > > In reference to [1] and the refactoring work, and after multiple > different > > tests, I came to realize that Tomcat clustering implementation is broken > > (both before and after the refactoring). The reasons are complex but they > > essentially have to do with setting up the right parameters for the > > cluster. So I suspect no one is using this feature? > > > > It is therefore, my recommendation to remove clustering from > > CatalinaContainer which would substantially reduce complexity of both the > > component configuration file and the accompanying source code which would > > shave off at least 120 lines of code. > > > > WDYT > > > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > > > On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < > > [hidden email]> > > wrote: > > > > > Hey folks, > > > > > > It was very painful and slow, but I finally got a working almost > > > full-rewrite of the CatalinaContainer. I need help with testing, > reviews, > > > and code improvements. > > > > > > Details are found in [1]. There are no functional changes, just a > > rewrite. > > > I appreciate all the help and feedback. > > > > > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > > > > > > Cheers, > > > > > > Taher Alkhateeb > > > > > > > > > > > > |
Okay, after a little bit of investigation I think I got what caused the
crashes on my machine: - I removed "cluster.registerManager(manager);" from the code base because it was commented out in the original code base - I commented out "<property name="mcast-bind-addr" value="192.168.2.1"/>" from ofbiz-component.xml because it was conflicting with my local LAN setup. After doing the above and testing the cluster, everything worked. Bug again, given that I'm treading in new territory, appreciate all the help. The new patch is available at https://issues.apache.org/jira/browse/OFBIZ-9392 Cheers, Taher Alkhateeb On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb <[hidden email] > wrote: > So the good thing is that the refactoring exercise surfaced this issue. > Based on feedback from Scott and Michael I suspect that the issue might be > perhaps minor such as a missing configuration or dependency that's causing > all these errors. > > I'll try to dig out the root cause and your help would be great to try and > zoom in on the problem preferably on the refactored code, but even finding > the root cause on existing code would greatly help. > > Cheers, > > Taher > > On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> > wrote: > >> The project I'm currently on is actively using clustering (session >> replication) and I know for certain this isn't the only project nor are we >> the only vendor to use it. I'm using an older version of OFBiz though so >> I >> can't confirm if the latest codebase has it working. It can be tricky to >> set up and does have a tendency to crash ofbiz during startup if you don't >> have it configured correctly. >> >> Definitely a -1 for removal. >> >> Regards >> Scott >> >> On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> >> wrote: >> >> > Hi Everyone, >> > >> > In reference to [1] and the refactoring work, and after multiple >> different >> > tests, I came to realize that Tomcat clustering implementation is broken >> > (both before and after the refactoring). The reasons are complex but >> they >> > essentially have to do with setting up the right parameters for the >> > cluster. So I suspect no one is using this feature? >> > >> > It is therefore, my recommendation to remove clustering from >> > CatalinaContainer which would substantially reduce complexity of both >> the >> > component configuration file and the accompanying source code which >> would >> > shave off at least 120 lines of code. >> > >> > WDYT >> > >> > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >> > >> > On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >> > [hidden email]> >> > wrote: >> > >> > > Hey folks, >> > > >> > > It was very painful and slow, but I finally got a working almost >> > > full-rewrite of the CatalinaContainer. I need help with testing, >> reviews, >> > > and code improvements. >> > > >> > > Details are found in [1]. There are no functional changes, just a >> > rewrite. >> > > I appreciate all the help and feedback. >> > > >> > > [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >> > > >> > > Cheers, >> > > >> > > Taher Alkhateeb >> > > >> > > >> > > >> > >> > |
Hi Taher,
Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: > Okay, after a little bit of investigation I think I got what caused the > crashes on my machine: > - I removed "cluster.registerManager(manager);" from the code base because > it was commented out in the original code base That explains why I didn't get these errors with the pre-refactoring setup :-) > - I commented out "<property name="mcast-bind-addr" value="192.168.2.1"/>" > from ofbiz-component.xml because it was conflicting with my local LAN setup. I did the same with the current trunk and now OFBiz is starting with enabled clustering. > After doing the above and testing the cluster, everything worked. Bug > again, given that I'm treading in new territory, appreciate all the help. Do we have some documentation on how to setup the cluster environment for testing, to have some kind of reference setup? It would help to make testing easier. I will try to review your patch in the next days. Regards, Michael > The new patch is available at > https://issues.apache.org/jira/browse/OFBIZ-9392 > > Cheers, > > Taher Alkhateeb > > On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb <[hidden email] >> wrote: >> So the good thing is that the refactoring exercise surfaced this issue. >> Based on feedback from Scott and Michael I suspect that the issue might be >> perhaps minor such as a missing configuration or dependency that's causing >> all these errors. >> >> I'll try to dig out the root cause and your help would be great to try and >> zoom in on the problem preferably on the refactored code, but even finding >> the root cause on existing code would greatly help. >> >> Cheers, >> >> Taher >> >> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> >> wrote: >> >>> The project I'm currently on is actively using clustering (session >>> replication) and I know for certain this isn't the only project nor are we >>> the only vendor to use it. I'm using an older version of OFBiz though so >>> I >>> can't confirm if the latest codebase has it working. It can be tricky to >>> set up and does have a tendency to crash ofbiz during startup if you don't >>> have it configured correctly. >>> >>> Definitely a -1 for removal. >>> >>> Regards >>> Scott >>> >>> On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> >>> wrote: >>> >>>> Hi Everyone, >>>> >>>> In reference to [1] and the refactoring work, and after multiple >>> different >>>> tests, I came to realize that Tomcat clustering implementation is broken >>>> (both before and after the refactoring). The reasons are complex but >>> they >>>> essentially have to do with setting up the right parameters for the >>>> cluster. So I suspect no one is using this feature? >>>> >>>> It is therefore, my recommendation to remove clustering from >>>> CatalinaContainer which would substantially reduce complexity of both >>> the >>>> component configuration file and the accompanying source code which >>> would >>>> shave off at least 120 lines of code. >>>> >>>> WDYT >>>> >>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>> [hidden email]> >>>> wrote: >>>> >>>>> Hey folks, >>>>> >>>>> It was very painful and slow, but I finally got a working almost >>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>> reviews, >>>>> and code improvements. >>>>> >>>>> Details are found in [1]. There are no functional changes, just a >>>> rewrite. >>>>> I appreciate all the help and feedback. >>>>> >>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> Cheers, >>>>> >>>>> Taher Alkhateeb >>>>> >>>>> >>>>> smime.p7s (5K) Download Attachment |
Hi Michael,
I think most of our documentation is outdated. So your best bet is [1]. I look forward to your feedback. @Scott you seem to have good experience with clustering, It would be really great if we can have your eyes on the code or some testing of the functionality. [1] https://tomcat.apache.org/tomcat-8.0-doc/cluster-howto.html Cheers, -- Taher Alkhateeb On Wed, Jun 14, 2017 at 11:07 PM, Michael Brohl <[hidden email]> wrote: > Hi Taher, > > Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: > >> Okay, after a little bit of investigation I think I got what caused the >> crashes on my machine: >> - I removed "cluster.registerManager(manager);" from the code base >> because >> it was commented out in the original code base >> > > That explains why I didn't get these errors with the pre-refactoring setup > :-) > > - I commented out "<property name="mcast-bind-addr" value="192.168.2.1"/>" >> from ofbiz-component.xml because it was conflicting with my local LAN >> setup. >> > > I did the same with the current trunk and now OFBiz is starting with > enabled clustering. > > After doing the above and testing the cluster, everything worked. Bug >> again, given that I'm treading in new territory, appreciate all the help. >> > > Do we have some documentation on how to setup the cluster environment for > testing, to have some kind of reference setup? > It would help to make testing easier. > > I will try to review your patch in the next days. > > Regards, > Michael > > > The new patch is available at >> https://issues.apache.org/jira/browse/OFBIZ-9392 >> >> Cheers, >> >> Taher Alkhateeb >> >> On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb < >> [hidden email] >> >>> wrote: >>> So the good thing is that the refactoring exercise surfaced this issue. >>> Based on feedback from Scott and Michael I suspect that the issue might >>> be >>> perhaps minor such as a missing configuration or dependency that's >>> causing >>> all these errors. >>> >>> I'll try to dig out the root cause and your help would be great to try >>> and >>> zoom in on the problem preferably on the refactored code, but even >>> finding >>> the root cause on existing code would greatly help. >>> >>> Cheers, >>> >>> Taher >>> >>> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> >>> wrote: >>> >>> The project I'm currently on is actively using clustering (session >>>> replication) and I know for certain this isn't the only project nor are >>>> we >>>> the only vendor to use it. I'm using an older version of OFBiz though >>>> so >>>> I >>>> can't confirm if the latest codebase has it working. It can be tricky >>>> to >>>> set up and does have a tendency to crash ofbiz during startup if you >>>> don't >>>> have it configured correctly. >>>> >>>> Definitely a -1 for removal. >>>> >>>> Regards >>>> Scott >>>> >>>> On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> >>>> wrote: >>>> >>>> Hi Everyone, >>>>> >>>>> In reference to [1] and the refactoring work, and after multiple >>>>> >>>> different >>>> >>>>> tests, I came to realize that Tomcat clustering implementation is >>>>> broken >>>>> (both before and after the refactoring). The reasons are complex but >>>>> >>>> they >>>> >>>>> essentially have to do with setting up the right parameters for the >>>>> cluster. So I suspect no one is using this feature? >>>>> >>>>> It is therefore, my recommendation to remove clustering from >>>>> CatalinaContainer which would substantially reduce complexity of both >>>>> >>>> the >>>> >>>>> component configuration file and the accompanying source code which >>>>> >>>> would >>>> >>>>> shave off at least 120 lines of code. >>>>> >>>>> WDYT >>>>> >>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>>> [hidden email]> >>>>> wrote: >>>>> >>>>> Hey folks, >>>>>> >>>>>> It was very painful and slow, but I finally got a working almost >>>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>>> >>>>> reviews, >>>> >>>>> and code improvements. >>>>>> >>>>>> Details are found in [1]. There are no functional changes, just a >>>>>> >>>>> rewrite. >>>>> >>>>>> I appreciate all the help and feedback. >>>>>> >>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>> >>>>>> Cheers, >>>>>> >>>>>> Taher Alkhateeb >>>>>> >>>>>> >>>>>> >>>>>> > > |
Hey Folks. I know everyone's probably preoccupied. It would be great to get
some feedback on OFBIZ-9392 though given how critical this piece of code is. I'd feel much better about committing after a few reviews. On Thu, Jun 15, 2017 at 9:22 PM, Taher Alkhateeb <[hidden email] > wrote: > Hi Michael, > > I think most of our documentation is outdated. So your best bet is [1]. I > look forward to your feedback. > > @Scott you seem to have good experience with clustering, It would be > really great if we can have your eyes on the code or some testing of the > functionality. > > [1] https://tomcat.apache.org/tomcat-8.0-doc/cluster-howto.html > > Cheers, > -- > Taher Alkhateeb > > On Wed, Jun 14, 2017 at 11:07 PM, Michael Brohl <[hidden email]> > wrote: > >> Hi Taher, >> >> Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: >> >>> Okay, after a little bit of investigation I think I got what caused the >>> crashes on my machine: >>> - I removed "cluster.registerManager(manager);" from the code base >>> because >>> it was commented out in the original code base >>> >> >> That explains why I didn't get these errors with the pre-refactoring >> setup :-) >> >> - I commented out "<property name="mcast-bind-addr" value="192.168.2.1"/>" >>> from ofbiz-component.xml because it was conflicting with my local LAN >>> setup. >>> >> >> I did the same with the current trunk and now OFBiz is starting with >> enabled clustering. >> >> After doing the above and testing the cluster, everything worked. Bug >>> again, given that I'm treading in new territory, appreciate all the help. >>> >> >> Do we have some documentation on how to setup the cluster environment for >> testing, to have some kind of reference setup? >> It would help to make testing easier. >> >> I will try to review your patch in the next days. >> >> Regards, >> Michael >> >> >> The new patch is available at >>> https://issues.apache.org/jira/browse/OFBIZ-9392 >>> >>> Cheers, >>> >>> Taher Alkhateeb >>> >>> On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb < >>> [hidden email] >>> >>>> wrote: >>>> So the good thing is that the refactoring exercise surfaced this issue. >>>> Based on feedback from Scott and Michael I suspect that the issue might >>>> be >>>> perhaps minor such as a missing configuration or dependency that's >>>> causing >>>> all these errors. >>>> >>>> I'll try to dig out the root cause and your help would be great to try >>>> and >>>> zoom in on the problem preferably on the refactored code, but even >>>> finding >>>> the root cause on existing code would greatly help. >>>> >>>> Cheers, >>>> >>>> Taher >>>> >>>> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> >>>> wrote: >>>> >>>> The project I'm currently on is actively using clustering (session >>>>> replication) and I know for certain this isn't the only project nor >>>>> are we >>>>> the only vendor to use it. I'm using an older version of OFBiz though >>>>> so >>>>> I >>>>> can't confirm if the latest codebase has it working. It can be tricky >>>>> to >>>>> set up and does have a tendency to crash ofbiz during startup if you >>>>> don't >>>>> have it configured correctly. >>>>> >>>>> Definitely a -1 for removal. >>>>> >>>>> Regards >>>>> Scott >>>>> >>>>> On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> >>>>> wrote: >>>>> >>>>> Hi Everyone, >>>>>> >>>>>> In reference to [1] and the refactoring work, and after multiple >>>>>> >>>>> different >>>>> >>>>>> tests, I came to realize that Tomcat clustering implementation is >>>>>> broken >>>>>> (both before and after the refactoring). The reasons are complex but >>>>>> >>>>> they >>>>> >>>>>> essentially have to do with setting up the right parameters for the >>>>>> cluster. So I suspect no one is using this feature? >>>>>> >>>>>> It is therefore, my recommendation to remove clustering from >>>>>> CatalinaContainer which would substantially reduce complexity of both >>>>>> >>>>> the >>>>> >>>>>> component configuration file and the accompanying source code which >>>>>> >>>>> would >>>>> >>>>>> shave off at least 120 lines of code. >>>>>> >>>>>> WDYT >>>>>> >>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>> >>>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>>>> [hidden email]> >>>>>> wrote: >>>>>> >>>>>> Hey folks, >>>>>>> >>>>>>> It was very painful and slow, but I finally got a working almost >>>>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>>>> >>>>>> reviews, >>>>> >>>>>> and code improvements. >>>>>>> >>>>>>> Details are found in [1]. There are no functional changes, just a >>>>>>> >>>>>> rewrite. >>>>>> >>>>>>> I appreciate all the help and feedback. >>>>>>> >>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>>> >>>>>>> Cheers, >>>>>>> >>>>>>> Taher Alkhateeb >>>>>>> >>>>>>> >>>>>>> >>>>>>> >> >> > |
Hi Taher,
The jira summary of changes sounds like a re-organization without any (intentional) changes in functionality. Personally I would just go ahead and commit it. Regards Scott On 20 June 2017 at 09:13, Taher Alkhateeb <[hidden email]> wrote: > Hey Folks. I know everyone's probably preoccupied. It would be great to get > some feedback on OFBIZ-9392 though given how critical this piece of code > is. I'd feel much better about committing after a few reviews. > > On Thu, Jun 15, 2017 at 9:22 PM, Taher Alkhateeb < > [hidden email] > > wrote: > > > Hi Michael, > > > > I think most of our documentation is outdated. So your best bet is [1]. I > > look forward to your feedback. > > > > @Scott you seem to have good experience with clustering, It would be > > really great if we can have your eyes on the code or some testing of the > > functionality. > > > > [1] https://tomcat.apache.org/tomcat-8.0-doc/cluster-howto.html > > > > Cheers, > > -- > > Taher Alkhateeb > > > > On Wed, Jun 14, 2017 at 11:07 PM, Michael Brohl < > [hidden email]> > > wrote: > > > >> Hi Taher, > >> > >> Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: > >> > >>> Okay, after a little bit of investigation I think I got what caused the > >>> crashes on my machine: > >>> - I removed "cluster.registerManager(manager);" from the code base > >>> because > >>> it was commented out in the original code base > >>> > >> > >> That explains why I didn't get these errors with the pre-refactoring > >> setup :-) > >> > >> - I commented out "<property name="mcast-bind-addr" > value="192.168.2.1"/>" > >>> from ofbiz-component.xml because it was conflicting with my local LAN > >>> setup. > >>> > >> > >> I did the same with the current trunk and now OFBiz is starting with > >> enabled clustering. > >> > >> After doing the above and testing the cluster, everything worked. Bug > >>> again, given that I'm treading in new territory, appreciate all the > help. > >>> > >> > >> Do we have some documentation on how to setup the cluster environment > for > >> testing, to have some kind of reference setup? > >> It would help to make testing easier. > >> > >> I will try to review your patch in the next days. > >> > >> Regards, > >> Michael > >> > >> > >> The new patch is available at > >>> https://issues.apache.org/jira/browse/OFBIZ-9392 > >>> > >>> Cheers, > >>> > >>> Taher Alkhateeb > >>> > >>> On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb < > >>> [hidden email] > >>> > >>>> wrote: > >>>> So the good thing is that the refactoring exercise surfaced this > issue. > >>>> Based on feedback from Scott and Michael I suspect that the issue > might > >>>> be > >>>> perhaps minor such as a missing configuration or dependency that's > >>>> causing > >>>> all these errors. > >>>> > >>>> I'll try to dig out the root cause and your help would be great to try > >>>> and > >>>> zoom in on the problem preferably on the refactored code, but even > >>>> finding > >>>> the root cause on existing code would greatly help. > >>>> > >>>> Cheers, > >>>> > >>>> Taher > >>>> > >>>> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> > >>>> wrote: > >>>> > >>>> The project I'm currently on is actively using clustering (session > >>>>> replication) and I know for certain this isn't the only project nor > >>>>> are we > >>>>> the only vendor to use it. I'm using an older version of OFBiz > though > >>>>> so > >>>>> I > >>>>> can't confirm if the latest codebase has it working. It can be > tricky > >>>>> to > >>>>> set up and does have a tendency to crash ofbiz during startup if you > >>>>> don't > >>>>> have it configured correctly. > >>>>> > >>>>> Definitely a -1 for removal. > >>>>> > >>>>> Regards > >>>>> Scott > >>>>> > >>>>> On 13 June 2017 at 08:57, Taher Alkhateeb < > [hidden email]> > >>>>> wrote: > >>>>> > >>>>> Hi Everyone, > >>>>>> > >>>>>> In reference to [1] and the refactoring work, and after multiple > >>>>>> > >>>>> different > >>>>> > >>>>>> tests, I came to realize that Tomcat clustering implementation is > >>>>>> broken > >>>>>> (both before and after the refactoring). The reasons are complex but > >>>>>> > >>>>> they > >>>>> > >>>>>> essentially have to do with setting up the right parameters for the > >>>>>> cluster. So I suspect no one is using this feature? > >>>>>> > >>>>>> It is therefore, my recommendation to remove clustering from > >>>>>> CatalinaContainer which would substantially reduce complexity of > both > >>>>>> > >>>>> the > >>>>> > >>>>>> component configuration file and the accompanying source code which > >>>>>> > >>>>> would > >>>>> > >>>>>> shave off at least 120 lines of code. > >>>>>> > >>>>>> WDYT > >>>>>> > >>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > >>>>>> > >>>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < > >>>>>> [hidden email]> > >>>>>> wrote: > >>>>>> > >>>>>> Hey folks, > >>>>>>> > >>>>>>> It was very painful and slow, but I finally got a working almost > >>>>>>> full-rewrite of the CatalinaContainer. I need help with testing, > >>>>>>> > >>>>>> reviews, > >>>>> > >>>>>> and code improvements. > >>>>>>> > >>>>>>> Details are found in [1]. There are no functional changes, just a > >>>>>>> > >>>>>> rewrite. > >>>>>> > >>>>>>> I appreciate all the help and feedback. > >>>>>>> > >>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 > >>>>>>> > >>>>>>> Cheers, > >>>>>>> > >>>>>>> Taher Alkhateeb > >>>>>>> > >>>>>>> > >>>>>>> > >>>>>>> > >> > >> > > > |
Administrator
|
In reply to this post by taher
Hi Taher,
I'll try to have a look today. Jacques Le 19/06/2017 à 23:13, Taher Alkhateeb a écrit : > Hey Folks. I know everyone's probably preoccupied. It would be great to get > some feedback on OFBIZ-9392 though given how critical this piece of code > is. I'd feel much better about committing after a few reviews. > > On Thu, Jun 15, 2017 at 9:22 PM, Taher Alkhateeb <[hidden email] >> wrote: >> Hi Michael, >> >> I think most of our documentation is outdated. So your best bet is [1]. I >> look forward to your feedback. >> >> @Scott you seem to have good experience with clustering, It would be >> really great if we can have your eyes on the code or some testing of the >> functionality. >> >> [1] https://tomcat.apache.org/tomcat-8.0-doc/cluster-howto.html >> >> Cheers, >> -- >> Taher Alkhateeb >> >> On Wed, Jun 14, 2017 at 11:07 PM, Michael Brohl <[hidden email]> >> wrote: >> >>> Hi Taher, >>> >>> Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: >>> >>>> Okay, after a little bit of investigation I think I got what caused the >>>> crashes on my machine: >>>> - I removed "cluster.registerManager(manager);" from the code base >>>> because >>>> it was commented out in the original code base >>>> >>> That explains why I didn't get these errors with the pre-refactoring >>> setup :-) >>> >>> - I commented out "<property name="mcast-bind-addr" value="192.168.2.1"/>" >>>> from ofbiz-component.xml because it was conflicting with my local LAN >>>> setup. >>>> >>> I did the same with the current trunk and now OFBiz is starting with >>> enabled clustering. >>> >>> After doing the above and testing the cluster, everything worked. Bug >>>> again, given that I'm treading in new territory, appreciate all the help. >>>> >>> Do we have some documentation on how to setup the cluster environment for >>> testing, to have some kind of reference setup? >>> It would help to make testing easier. >>> >>> I will try to review your patch in the next days. >>> >>> Regards, >>> Michael >>> >>> >>> The new patch is available at >>>> https://issues.apache.org/jira/browse/OFBIZ-9392 >>>> >>>> Cheers, >>>> >>>> Taher Alkhateeb >>>> >>>> On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb < >>>> [hidden email] >>>> >>>>> wrote: >>>>> So the good thing is that the refactoring exercise surfaced this issue. >>>>> Based on feedback from Scott and Michael I suspect that the issue might >>>>> be >>>>> perhaps minor such as a missing configuration or dependency that's >>>>> causing >>>>> all these errors. >>>>> >>>>> I'll try to dig out the root cause and your help would be great to try >>>>> and >>>>> zoom in on the problem preferably on the refactored code, but even >>>>> finding >>>>> the root cause on existing code would greatly help. >>>>> >>>>> Cheers, >>>>> >>>>> Taher >>>>> >>>>> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> >>>>> wrote: >>>>> >>>>> The project I'm currently on is actively using clustering (session >>>>>> replication) and I know for certain this isn't the only project nor >>>>>> are we >>>>>> the only vendor to use it. I'm using an older version of OFBiz though >>>>>> so >>>>>> I >>>>>> can't confirm if the latest codebase has it working. It can be tricky >>>>>> to >>>>>> set up and does have a tendency to crash ofbiz during startup if you >>>>>> don't >>>>>> have it configured correctly. >>>>>> >>>>>> Definitely a -1 for removal. >>>>>> >>>>>> Regards >>>>>> Scott >>>>>> >>>>>> On 13 June 2017 at 08:57, Taher Alkhateeb <[hidden email]> >>>>>> wrote: >>>>>> >>>>>> Hi Everyone, >>>>>>> In reference to [1] and the refactoring work, and after multiple >>>>>>> >>>>>> different >>>>>> >>>>>>> tests, I came to realize that Tomcat clustering implementation is >>>>>>> broken >>>>>>> (both before and after the refactoring). The reasons are complex but >>>>>>> >>>>>> they >>>>>> >>>>>>> essentially have to do with setting up the right parameters for the >>>>>>> cluster. So I suspect no one is using this feature? >>>>>>> >>>>>>> It is therefore, my recommendation to remove clustering from >>>>>>> CatalinaContainer which would substantially reduce complexity of both >>>>>>> >>>>>> the >>>>>> >>>>>>> component configuration file and the accompanying source code which >>>>>>> >>>>>> would >>>>>> >>>>>>> shave off at least 120 lines of code. >>>>>>> >>>>>>> WDYT >>>>>>> >>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>>> >>>>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>>>>> [hidden email]> >>>>>>> wrote: >>>>>>> >>>>>>> Hey folks, >>>>>>>> It was very painful and slow, but I finally got a working almost >>>>>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>>>>> >>>>>>> reviews, >>>>>>> and code improvements. >>>>>>>> Details are found in [1]. There are no functional changes, just a >>>>>>>> >>>>>>> rewrite. >>>>>>> >>>>>>>> I appreciate all the help and feedback. >>>>>>>> >>>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>>>> >>>>>>>> Cheers, >>>>>>>> >>>>>>>> Taher Alkhateeb >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>> |
In reply to this post by Scott Gray-3
Hi, I agree with Scott, I check your code and I see no reason to don't
commit it Nicolas Le 20/06/2017 à 07:47, Scott Gray a écrit : > Hi Taher, > > The jira summary of changes sounds like a re-organization without any > (intentional) changes in functionality. Personally I would just go ahead > and commit it. > > Regards > Scott > > On 20 June 2017 at 09:13, Taher Alkhateeb <[hidden email]> > wrote: > >> Hey Folks. I know everyone's probably preoccupied. It would be great to get >> some feedback on OFBIZ-9392 though given how critical this piece of code >> is. I'd feel much better about committing after a few reviews. >> >> On Thu, Jun 15, 2017 at 9:22 PM, Taher Alkhateeb < >> [hidden email] >>> wrote: >>> Hi Michael, >>> >>> I think most of our documentation is outdated. So your best bet is [1]. I >>> look forward to your feedback. >>> >>> @Scott you seem to have good experience with clustering, It would be >>> really great if we can have your eyes on the code or some testing of the >>> functionality. >>> >>> [1] https://tomcat.apache.org/tomcat-8.0-doc/cluster-howto.html >>> >>> Cheers, >>> -- >>> Taher Alkhateeb >>> >>> On Wed, Jun 14, 2017 at 11:07 PM, Michael Brohl < >> [hidden email]> >>> wrote: >>> >>>> Hi Taher, >>>> >>>> Am 14.06.17 um 19:56 schrieb Taher Alkhateeb: >>>> >>>>> Okay, after a little bit of investigation I think I got what caused the >>>>> crashes on my machine: >>>>> - I removed "cluster.registerManager(manager);" from the code base >>>>> because >>>>> it was commented out in the original code base >>>>> >>>> That explains why I didn't get these errors with the pre-refactoring >>>> setup :-) >>>> >>>> - I commented out "<property name="mcast-bind-addr" >> value="192.168.2.1"/>" >>>>> from ofbiz-component.xml because it was conflicting with my local LAN >>>>> setup. >>>>> >>>> I did the same with the current trunk and now OFBiz is starting with >>>> enabled clustering. >>>> >>>> After doing the above and testing the cluster, everything worked. Bug >>>>> again, given that I'm treading in new territory, appreciate all the >> help. >>>> Do we have some documentation on how to setup the cluster environment >> for >>>> testing, to have some kind of reference setup? >>>> It would help to make testing easier. >>>> >>>> I will try to review your patch in the next days. >>>> >>>> Regards, >>>> Michael >>>> >>>> >>>> The new patch is available at >>>>> https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>> >>>>> Cheers, >>>>> >>>>> Taher Alkhateeb >>>>> >>>>> On Tue, Jun 13, 2017 at 9:17 AM, Taher Alkhateeb < >>>>> [hidden email] >>>>> >>>>>> wrote: >>>>>> So the good thing is that the refactoring exercise surfaced this >> issue. >>>>>> Based on feedback from Scott and Michael I suspect that the issue >> might >>>>>> be >>>>>> perhaps minor such as a missing configuration or dependency that's >>>>>> causing >>>>>> all these errors. >>>>>> >>>>>> I'll try to dig out the root cause and your help would be great to try >>>>>> and >>>>>> zoom in on the problem preferably on the refactored code, but even >>>>>> finding >>>>>> the root cause on existing code would greatly help. >>>>>> >>>>>> Cheers, >>>>>> >>>>>> Taher >>>>>> >>>>>> On Jun 13, 2017 4:16 AM, "Scott Gray" <[hidden email]> >>>>>> wrote: >>>>>> >>>>>> The project I'm currently on is actively using clustering (session >>>>>>> replication) and I know for certain this isn't the only project nor >>>>>>> are we >>>>>>> the only vendor to use it. I'm using an older version of OFBiz >> though >>>>>>> so >>>>>>> I >>>>>>> can't confirm if the latest codebase has it working. It can be >> tricky >>>>>>> to >>>>>>> set up and does have a tendency to crash ofbiz during startup if you >>>>>>> don't >>>>>>> have it configured correctly. >>>>>>> >>>>>>> Definitely a -1 for removal. >>>>>>> >>>>>>> Regards >>>>>>> Scott >>>>>>> >>>>>>> On 13 June 2017 at 08:57, Taher Alkhateeb < >> [hidden email]> >>>>>>> wrote: >>>>>>> >>>>>>> Hi Everyone, >>>>>>>> In reference to [1] and the refactoring work, and after multiple >>>>>>>> >>>>>>> different >>>>>>> >>>>>>>> tests, I came to realize that Tomcat clustering implementation is >>>>>>>> broken >>>>>>>> (both before and after the refactoring). The reasons are complex but >>>>>>>> >>>>>>> they >>>>>>> >>>>>>>> essentially have to do with setting up the right parameters for the >>>>>>>> cluster. So I suspect no one is using this feature? >>>>>>>> >>>>>>>> It is therefore, my recommendation to remove clustering from >>>>>>>> CatalinaContainer which would substantially reduce complexity of >> both >>>>>>> the >>>>>>> >>>>>>>> component configuration file and the accompanying source code which >>>>>>>> >>>>>>> would >>>>>>> >>>>>>>> shave off at least 120 lines of code. >>>>>>>> >>>>>>>> WDYT >>>>>>>> >>>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>>>> >>>>>>>> On Thu, Jun 8, 2017 at 3:38 PM, Taher Alkhateeb < >>>>>>>> [hidden email]> >>>>>>>> wrote: >>>>>>>> >>>>>>>> Hey folks, >>>>>>>>> It was very painful and slow, but I finally got a working almost >>>>>>>>> full-rewrite of the CatalinaContainer. I need help with testing, >>>>>>>>> >>>>>>>> reviews, >>>>>>>> and code improvements. >>>>>>>>> Details are found in [1]. There are no functional changes, just a >>>>>>>>> >>>>>>>> rewrite. >>>>>>>> >>>>>>>>> I appreciate all the help and feedback. >>>>>>>>> >>>>>>>>> [1] https://issues.apache.org/jira/browse/OFBIZ-9392 >>>>>>>>> >>>>>>>>> Cheers, >>>>>>>>> >>>>>>>>> Taher Alkhateeb >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> >>>> |
Free forum by Nabble | Edit this page |