uPortal IRC Logs-2011-10-10

Owned by Apereo Infrastructure
Last updated: Oct 10, 2011Version comment

[09:41:05 CDT(-0500)] <Arvids> Finally getting used to fluid and now I can translate portlet definitions... it feels so good (smile)

[09:41:56 CDT(-0500)] <Arvids> still feel like someone who's more used to building fancy UI, should later review my code

[09:49:52 CDT(-0500)] <Arvids> EricDalquist: how's going on with migrating to GIT?

[10:58:15 CDT(-0500)] <EricDalquist> hi Arvids

[10:58:33 CDT(-0500)] <EricDalquist> working on the plans right now actually

[10:58:42 CDT(-0500)] <EricDalquist> I'll have a full timeline proposal posted soon

[10:58:47 CDT(-0500)] <EricDalquist> there will be ample notice (smile)

[13:22:09 CDT(-0500)] <athena> anyone seen this error before? Caused by: java.lang.IllegalArgumentException: interface org.hibernate.engine.jdbc.ClobImplementer is not visible from class loader

[13:22:24 CDT(-0500)] <EricDalquist> neat ... no

[13:22:28 CDT(-0500)] <EricDalquist> do you have a stack trace?

[13:22:35 CDT(-0500)] <EricDalquist> because I have a sneaking fear ...

[13:22:49 CDT(-0500)] <athena> it's in a portlet, not uportal

[13:23:07 CDT(-0500)] <EricDalquist> could still be a uPortal/portlet classloader issue

[13:23:25 CDT(-0500)] <EricDalquist> uPortal is not as good about classloader demarcation as it should beo

[13:23:58 CDT(-0500)] <athena> http://pastebin.com/wbcieS2F

[13:24:27 CDT(-0500)] * EricDalquist hates that pastebin linewraps stack traces

[13:24:29 CDT(-0500)] <athena> er, think that pasted kinda weird.

[13:24:35 CDT(-0500)] <EricDalquist> https://gist.github.com

[13:24:37 CDT(-0500)] <EricDalquist> try that

[13:24:41 CDT(-0500)] <EricDalquist> login with your github info

[13:24:50 CDT(-0500)] <athena> think part of it was how i copied it in the first place

[13:24:53 CDT(-0500)] <EricDalquist> ah

[13:25:56 CDT(-0500)] <athena> https://gist.github.com/1276097

[13:25:58 CDT(-0500)] <athena> there we go

[13:28:23 CDT(-0500)] <athena> tried updating to the version of hibernate uportal is using, didn't help (was using a really old vesion)

[13:28:32 CDT(-0500)] <athena> hmm!

[13:28:38 CDT(-0500)] <athena> now seeing two versions of cglib

[13:28:41 CDT(-0500)] <athena> that could be a problem

[13:28:44 CDT(-0500)] <EricDalquist> yeah

[13:28:52 CDT(-0500)] <athena> think the version of commons-dbcp matters at all?

[13:28:53 CDT(-0500)] <EricDalquist> my other concern is how pluto and uPortal handle classloaders

[13:28:56 CDT(-0500)] <EricDalquist> probably not

[13:29:04 CDT(-0500)] <EricDalquist> I've always just kinda hoped it all worked

[13:29:19 CDT(-0500)] <EricDalquist> but what we really need to do is go through and audit pluto and uPortal's code

[13:29:38 CDT(-0500)] <EricDalquist> and make sure that we're switching the thread's contextClassloader correctly

[13:34:28 CDT(-0500)] <EricDalquist> so yeah this is a portlet

[13:34:34 CDT(-0500)] <EricDalquist> but it is storing preferences

[13:34:39 CDT(-0500)] <EricDalquist> and the root of the exception is in uPortal

[13:38:06 CDT(-0500)] <EricDalquist> yeah I think this is a classloader issue on uPortal's side

[13:39:33 CDT(-0500)] <EricDalquist> athena: not sure how much work it would be but can you try applying this patch: https://gist.github.com/1276132

[13:39:39 CDT(-0500)] <EricDalquist> to uPortal and then seeing if that fixes the bug?

[13:43:15 CDT(-0500)] <athena> sure, if removing cglib doesnt' fix it i'll try that next

[13:43:21 CDT(-0500)] <athena> it only seems to happen on psql

[13:43:46 CDT(-0500)] <EricDalquist> I'm 99% sure that it is uPortal's fault

[13:43:57 CDT(-0500)] <athena> eh-oh (smile)

[13:44:26 CDT(-0500)] <athena> but yeah i'll totally try that out

[13:44:31 CDT(-0500)] <EricDalquist> what we really need to do is figure out a way to automatically set the thread ContextClassLoader back to uPortal's classloader any time a portlet/pluto makes a call back into uPortal

[13:44:44 CDT(-0500)] <EricDalquist> and then reset the loader back to the portlets when the stack unwinds

[13:45:09 CDT(-0500)] <EricDalquist> I think tomcat is taking care of switching to the portlet's classloader for us in the cross-context dispatch

[13:45:39 CDT(-0500)] <athena> gotcha

[13:45:53 CDT(-0500)] <athena> sort of weird that only this one portlet is triggering the error

[13:46:03 CDT(-0500)] <EricDalquist> its likely a race condition as well

[13:46:20 CDT(-0500)] <EricDalquist> since if that class has been loaded correctly before it should all just work

[13:46:27 CDT(-0500)] <EricDalquist> (tongue)

[13:46:28 CDT(-0500)] <athena> ah

[13:46:48 CDT(-0500)] <EricDalquist> threading + multiple classloaders + reflection

[13:46:51 CDT(-0500)] <athena> (tongue)

[13:46:51 CDT(-0500)] <EricDalquist> == FUN

[13:46:52 CDT(-0500)] <athena> ick!

[13:47:05 CDT(-0500)] <athena> fun for some super-nerdy and bored individual? (smile)

[13:47:10 CDT(-0500)] <EricDalquist> lol

[13:47:25 CDT(-0500)] <athena> i'm taking compilers this quarter

[13:47:35 CDT(-0500)] <athena> spent several hours creating regexes and DFAs from a bar yesterday

[13:47:46 CDT(-0500)] <athena> i think i would have enjoyed watching football more

[13:48:19 CDT(-0500)] <EricDalquist> hah

[13:48:48 CDT(-0500)] <EricDalquist> I kinda regret not taking a compilers course

[13:48:55 CDT(-0500)] <EricDalquist> no interest in actually writing them

[13:49:03 CDT(-0500)] <athena> we'll see if i wind up regretting taking this one

[13:49:04 CDT(-0500)] <EricDalquist> just think it would be a useful bit of background knowledge

[13:49:06 CDT(-0500)] <athena> yeah

[13:49:16 CDT(-0500)] <athena> and relevant for writing small parsers and things

[13:49:47 CDT(-0500)] <EricDalquist> yup

[13:54:42 CDT(-0500)] <b-rock> Greetings uPortal devs. I've got a build of uPortal4.0.1 running after an initdb with the default included data. When I log in as myself, I get an error and see this WARN in the logs: http://pastebin.com/K3xQDE3j

[13:55:10 CDT(-0500)] <EricDalquist> do you have any content when you first log in?

[13:55:30 CDT(-0500)] <b-rock> there is, our default tab has a bunch of stuff in it

[13:55:44 CDT(-0500)] <b-rock> do I just need to import my layout?

[13:56:06 CDT(-0500)] <EricDalquist> huh, the warning is fairly harmless

[13:56:08 CDT(-0500)] <b-rock> I'm trying to figure out the least amount of data I need to import from a previous build to get in

[13:56:52 CDT(-0500)] <EricDalquist> the portal is trying to figure out the default tab for you and can't

[13:57:40 CDT(-0500)] <EricDalquist> it is running the XPath: /layout/folder/folder[@type='regular' and @hidden!='true'][$defaultTab]/@ID

[13:57:44 CDT(-0500)] <EricDalquist> against your layout

[13:57:58 CDT(-0500)] <EricDalquist> and according to the log message $defaultTab is 1

[13:58:07 CDT(-0500)] <EricDalquist> so for some reason it isn't finding that tab

[13:58:09 CDT(-0500)] <b-rock> here is a java trace for it http://pastebin.com/g6KJNw84

[13:58:43 CDT(-0500)] <EricDalquist> oh so you're getting an error?

[13:58:52 CDT(-0500)] <EricDalquist> I missed that bit (tongue)

[13:59:04 CDT(-0500)] <EricDalquist> can you go into your log4j.properties and add:

[13:59:04 CDT(-0500)] <EricDalquist> log4j.logger.org.jasig.portal.rendering.LoggingStAXComponent.POST_LAYOUT=DEBUG, R

[13:59:04 CDT(-0500)] <EricDalquist> log4j.additivity.org.jasig.portal.rendering.LoggingStAXComponent.POST_LAYOUT=false

[13:59:06 CDT(-0500)] <b-rock> yeah I just get the defautl error page after being logged in. the portal page does not render

[13:59:18 CDT(-0500)] <EricDalquist> that should get you a nice pretty-printed XML document in your output

[13:59:20 CDT(-0500)] <EricDalquist> which is your layout

[13:59:26 CDT(-0500)] <b-rock> oh ok sure...

[13:59:27 CDT(-0500)] <EricDalquist> can you pastebin that

[13:59:36 CDT(-0500)] <b-rock> yep. just a moment...

[13:59:59 CDT(-0500)] <EricDalquist> my bet: you don't actually have any tabs and the url processing code barfs for that case

[14:04:10 CDT(-0500)] <b-rock> shoot. I need to go to meeting. I'll try to post the xml this afternoon. Thanks for your help EricDalquist.

[14:04:17 CDT(-0500)] <EricDalquist> ok

[14:39:04 CDT(-0500)] <athena1> EricDalquist: think that fixed it

[14:39:08 CDT(-0500)] <EricDalquist> yay

[14:39:09 CDT(-0500)] <EricDalquist> ok

[14:39:11 CDT(-0500)] <EricDalquist> I'll commit it

[14:39:15 CDT(-0500)] <athena1> awesome (smile)

[14:42:53 CDT(-0500)] <athena1> hmm

[14:43:01 CDT(-0500)] <athena1> this ProxyView class doesn't seem to be behaving anymore

[14:43:03 CDT(-0500)] <athena1> used to work

[14:51:09 CDT(-0500)] <athena1> so EricDalquist - https://source.jasig.org/sandbox/AjaxPortletSupport/trunk/src/main/java/org/jasig/web/view/mvc/ProxyView.java

[14:51:25 CDT(-0500)] <athena1> code seems to work locally, but not on more production servers

[14:51:28 CDT(-0500)] <athena1> any thoughts?

[14:51:47 CDT(-0500)] <athena1> just sends back a blank response

[14:54:29 CDT(-0500)] <EricDalquist> do you get any headers back?

[14:54:53 CDT(-0500)] <EricDalquist> or anything in the logs?

[14:55:07 CDT(-0500)] <EricDalquist> since it catches the exceptions it will always reutrn a 200

[14:55:08 CDT(-0500)] <EricDalquist> which isn't that good

[14:55:17 CDT(-0500)] <EricDalquist> it should really return a 503 if the proxying fails

[14:55:21 CDT(-0500)] <athena1> yeah, it's kind of a crappy class in general

[14:55:23 CDT(-0500)] <EricDalquist> would at least be easier to debug

[14:55:28 CDT(-0500)] <athena1> but no exceptions in the logs

[14:55:31 CDT(-0500)] <EricDalquist> those are my only ideas

[14:55:33 CDT(-0500)] <athena1> and the two debug statements print out ok

[14:55:49 CDT(-0500)] <athena1> and tried dup'ing the code and printing out the resutls of the http request and that does return content

[14:56:01 CDT(-0500)] <athena1> so it seems like it's just failing to actually copy it to the output

[14:56:03 CDT(-0500)] <athena1> which is really weird

[14:56:08 CDT(-0500)] <EricDalquist> yes

[14:56:24 CDT(-0500)] <athena1> and even weirder that it works locally

[14:56:34 CDT(-0500)] <EricDalquist> yes

[14:56:41 CDT(-0500)] <athena1> not sure if adding apache or something is doing something strange

[14:56:50 CDT(-0500)] <athena1> but wanted to make sure there wasn't something really obvious that i missed :/

[14:59:36 CDT(-0500)] <EricDalquist> not that I can see :/

[14:59:47 CDT(-0500)] <athena1> ok, thanks at least

[14:59:56 CDT(-0500)] <athena1> maybe i'll see if cleaning up that code makes any difference

[15:26:12 CDT(-0500)] <b-rock> Hi EricDalquist. I've got some more debug info in the logs for the pastebin above, but no rendering of the layout in xml. http://pastebin.com/BZh81AKk maybe the error happens before the xml can be logged.

[15:26:26 CDT(-0500)] <EricDalquist> hrm you're right

[15:26:46 CDT(-0500)] <EricDalquist> can you create a jira issue with the stack trace from: http://pastebin.com/g6KJNw84

[15:26:52 CDT(-0500)] <EricDalquist> (prefferablly not linewrapped (smile) )

[15:27:12 CDT(-0500)] <b-rock> sure.

[15:33:11 CDT(-0500)] <b-rock> ok I created this https://issues.jasig.org/browse/UP-3194. As I workaround, do you think I should just try it import the user layouts from our current 3.2.5-patches?

[15:33:41 CDT(-0500)] <EricDalquist> double check and make sure your user qualifies for at least one DLM managed fragment

[15:33:42 CDT(-0500)] <b-rock> thanks for your help BTW (smile)

[15:34:21 CDT(-0500)] <EricDalquist> I really think that the core of the problem is that you're ending up with no tabs

[15:35:34 CDT(-0500)] <b-rock> hmm. ok. I do notice that the unauthenticated guest view does render.

[15:38:09 CDT(-0500)] <EricDalquist> and it has at least one tab right?

[15:41:59 CDT(-0500)] <b-rock> yes it has the Welcome tab but there is no content/portlets displaying

[15:42:06 CDT(-0500)] <EricDalquist> right

[15:42:08 CDT(-0500)] <EricDalquist> but it has a tab

[15:42:12 CDT(-0500)] <b-rock> it does

[15:42:26 CDT(-0500)] <EricDalquist> my guess is that your account does not qualify for any DLM fragments

[15:42:29 CDT(-0500)] <EricDalquist> and so has no tabs

[15:42:52 CDT(-0500)] <EricDalquist> and there is apparently a bug in the new URL syntax handling code when a layout has no tabs

[15:43:10 CDT(-0500)] <b-rock> double checking the logs now...

[15:49:59 CDT(-0500)] <b-rock> so I have this from the FragmentDefinition.isApplicible. http://pastebin.com/UZ20Rceb

[15:50:22 CDT(-0500)] <EricDalquist> yes and we established that your guest user is working

[15:50:28 CDT(-0500)] <EricDalquist> the question is your billbrown user

[15:50:30 CDT(-0500)] <EricDalquist> which isn't working

[15:50:40 CDT(-0500)] <EricDalquist> either that user has no applicable fragments

[15:50:47 CDT(-0500)] <EricDalquist> or the fragments that are applicable have no content

[15:51:09 CDT(-0500)] <b-rock> oh ok. I'll dig that up...

[15:51:22 CDT(-0500)] <EricDalquist> I'm heading home

[15:51:27 CDT(-0500)] <EricDalquist> we can catch up tomorrow

[15:51:49 CDT(-0500)] <b-rock> sure np. thanks for your help. could be that all the portlets on the fragment are broken (incompatible portlet 1.0 version)

[15:52:10 CDT(-0500)] <EricDalquist> that isn't the case

[15:52:21 CDT(-0500)] <EricDalquist> and 1.0 portlets work just fine in 2.0

[15:52:27 CDT(-0500)] <EricDalquist> the spec is 100% backwards compatible

[15:52:40 CDT(-0500)] <EricDalquist> only 1 or 2 portlets that ship with up4 are 2.0 portlets

[15:52:42 CDT(-0500)] <EricDalquist> the rest are still 1.0