I don't remember that conversation, but I will try to track it down. Thx...
On Mar 9, 2017 6:39 PM, "Rafael Weingärtner" <rafaelweingart...@gmail.com> wrote: > I recall a thread discussing some call home functionalities a long time > ago; so, we could get some information about ACS usage. Does anyone know > how that thread ended? > > The use o ACS plugin could be something trackable by a call home feature. > > On Thu, Mar 9, 2017 at 5:55 PM, Will Stevens <wstev...@cloudops.com> > wrote: > > > We have had similar conversations many times recently. Unfortunately we > > have no way to track IF people are using different plugins, so it makes > it > > really hard to know if people will expect it to be there if they > upgrade... > > > > If anyone has ideas for how we can potentially find a way to track that, > we > > should probably start a thread around that. > > > > *Will STEVENS* > > Lead Developer > > > > <https://goo.gl/NYZ8KK> > > > > On Thu, Mar 9, 2017 at 2:36 PM, Rafael Weingärtner < > > rafaelweingart...@gmail.com> wrote: > > > > > Agree with you, but I think a nice talk with people using it to ask for > > > help might be a good idea. > > > > > > Sometimes they are not aware of these situations. > > > > > > On Thu, Mar 9, 2017 at 2:23 PM, Daan Hoogland <daan.hoogl...@gmail.com > > > > > wrote: > > > > > > > If we have to lay a burdon of fixing before release on people that > > > > don't use it, i would say no. You use it, you maintain it. I am happy > > > > to charge money to whoever wants to not maintain what they use. > > > > > > > > On Thu, Mar 9, 2017 at 7:11 PM, Rafael Weingärtner > > > > <rafaelweingart...@gmail.com> wrote: > > > > > Daan, before removing anything, I think we should check if there > are > > > > people > > > > > using it, right? > > > > > > > > > > On Thu, Mar 9, 2017 at 11:08 AM, Daan Hoogland < > > > daan.hoogl...@gmail.com> > > > > > wrote: > > > > > > > > > >> I was about to reply along those lines. As you brought it up, we > are > > > > >> now considering it. If the fix is easy I'd say let it stay till > the > > > > >> next problem but it is ot the first time mido bugs us. > > > > >> > > > > >> On Thu, Mar 9, 2017 at 5:07 PM, Simon Weller <swel...@ena.com> > > wrote: > > > > >> > So this brings up a good discussion point. As Jeff points out, > the > > > > >> Midonet plugin hasn't been actively supported for almost 5 years. > At > > > > what > > > > >> point do we consider retiring unsupported plugins? > > > > >> > > > > > >> > > > > > >> > - Si > > > > >> > > > > > >> > > > > > >> > ________________________________ > > > > >> > From: Jeff Hair <j...@greenqloud.com> > > > > >> > Sent: Thursday, March 9, 2017 9:43 AM > > > > >> > To: dev@cloudstack.apache.org > > > > >> > Subject: Re: midonet-client and Guava dependency conflict > > > > >> > > > > > >> > After doing some more digging, I have confirmed the following: > > > > >> > > > > > >> > - The midonet plugin is using the Maven Shade plugin to put a > > > > bunch of > > > > >> > dependencies into itself. > > > > >> > - The plugin hosted in this repository was last updated in > > 2013. > > > > >> > - Most importantly: removing all the guava stuff out of the > > > midonet > > > > >> > plugin fixes this issue. > > > > >> > > > > > >> > I have not had any success in applying > > > > >> > https://github.com/openwide-java/tomcat-classloader-ordered to > > get > > > > >> Tomcat > > > > >> > [https://avatars1.githubusercontent.com/u/1385131?v=3&s=400 > > > ]<https:// > > > > >> github.com/openwide-java/tomcat-classloader-ordered> > > > > >> > > > > > >> > GitHub - openwide-java/tomcat-classloader-ordered: A ...< > > > > >> https://github.com/openwide-java/tomcat-classloader-ordered> > > > > >> > github.com > > > > >> > README.md tomcat-classloader-ordered. A classloader for Apache > > > Tomcat > > > > 8 > > > > >> which loads the jars of WEB-INF lib in alphabetical order. Prior > to > > > > version > > > > >> 8, Apache Tomcat ... > > > > >> > > > > > >> > > > > > >> > > > > > >> > to load its jars in alphabetical order, for whatever reason. I > > tried > > > > >> > putting the Loader in various context definition locations, but > it > > > > >> refuses > > > > >> > to work. Any ideas? > > > > >> > > > > > >> > Jeff > > > > >> > > > > > >> > > > > > >> > On Thu, Mar 9, 2017 at 1:43 PM, Jeff Hair <j...@greenqloud.com> > > > > wrote: > > > > >> > > > > > >> >> Hi, > > > > >> >> > > > > >> >> I'm deploying 4.9.2.0 (not the vanilla version, but rather an > > > > upgraded > > > > >> >> version of our fork) on Tomcat 8. Management server startup > fails > > > > with > > > > >> the > > > > >> >> following error: > > > > >> >> > > > > >> >> java.lang.IncompatibleClassChangeError: Found interface > > > > >> >> com.google.common.base.Equivalence, but class was expected > > > > >> >> > > > > >> >> I've traced this down to the OutOfBandServiceManagerImpl. More > > > > >> >> specifically, when it tries to build the hostAlertCache using > > > Guava's > > > > >> >> CacheBuilder. Deep in Guava, it's calling an "identity()" > method > > on > > > > the > > > > >> >> Equivalence class. All of the Guava classes are coming from > > > > guava-19.0 > > > > >> >> except for com/google/common/base/Equivalence.class. The > > > Equivalence > > > > >> >> class is being loaded from the midonet jar for some reason, and > > > that > > > > >> >> version does not have the method needed. Thus, the error. > > > > >> >> > > > > >> >> This is because Tomcat apparently does not load jars in > > > alphabetical > > > > >> order > > > > >> >> anymore, starting with version 8. An open ticket for them to > fix > > > > this is > > > > >> >> here: https://bz.apache.org/bugzilla/show_bug.cgi?id=57129 > > > > >> > 57129 – Regression. Load WEB-INF/lib jarfiles in ...< > > > > >> https://bz.apache.org/bugzilla/show_bug.cgi?id=57129> > > > > >> > bz.apache.org > > > > >> > ASF Bugzilla – Bug 57129 Regression. Load WEB-INF/lib jarfiles > in > > > > >> alphabetical order Last modified: 2016-03-17 09:59:50 UTC > > > > >> > > > > > >> > > > > > >> > > > > > >> >> > > > > >> >> It could be possible to "fix" this by using a custom > ClassLoader > > to > > > > >> force > > > > >> >> Tomcat to load things alphabetically (testing that right > now--and > > > not > > > > >> >> really succeeding), but the proper fix is to have the midonet > > > client > > > > >> not be > > > > >> >> packaging guava with itself. Does anyone know why this is? > > > > >> >> > > > > >> >> Jeff > > > > >> >> > > > > >> > > > > >> > > > > >> > > > > >> -- > > > > >> Daan > > > > >> > > > > > > > > > > > > > > > > > > > > -- > > > > > Rafael Weingärtner > > > > > > > > > > > > > > > > -- > > > > Daan > > > > > > > > > > > > > > > > -- > > > Rafael Weingärtner > > > > > > > > > -- > Rafael Weingärtner >
