I'm trying to get my MG 3.2/Coldspring/Transfer app up to snuff so I can
begin doing true continuous deployment. I have refactored session objects
in order to persist across service layer and framework restarts
successfully. I'm running very simple tests where I restart my app in one
browser and click around in another. I can pretty easily get Model-Glue
(Controllers) to throw nonsense errors while it's reinitializing (meaning,
reloading a moment later, the error doesn't occur, and it's clear by the
error that it's an in-progress error where the controller is not fully
initialized yet). I also run into issues with Transfer emitting
CacheShutdown errors although I think I have a fix for that.
How do I re-init my app's service layer and model-glue apps without letting
a user make a request against an uninitialized framework? Here's my setup:
* parent beanfactory setup as application.cs
* one MG app and two MG sub-apps in sub directories, call them "public",
"subapp1" and "subapp2"
* MG 3.2
* Transfer ORM with EhCache
* My trigger "reinitfull' which restarts my beanfactory is NOT the same key
as Model-Glue's reload; I do that based on a suggestion by Sean Corfield to
just delete the key from the application scope which forces all three of my
MG apps to reload simultaneously.
My Application.cfc onApplicationStart() looks like:
<!--- create coldspring bean factory --->
<cfset bf = createObject("component",
"coldspring.beans.DefaultXmlBeanFactory").init() />
<cfset
bf.loadBeansFromXmlFile(expandPath("/PUKKA_MAIN_MAP/config/globalbeans.xml"),
true) />
<cfset application.cs = bf />
My OnRequestStart() looks like:
<cfif NOT structKeyExists(application, "cs") OR structKeyExists(URL,
"reinitfull")>
<cflock name="applicationReinit" type="exclusive" timeout="60">
<cfif NOT structKeyExists(application, "cs") OR
structKeyExists(URL, "reinitfull")>
<cfset msg = "Full init called" />
<cfinvoke component="cfide.adminapi.administrator"
method="login">
<cfinvokeargument name="adminPassword" value="somePassword"
/>
</cfinvoke>
<!--- clear template cache --->
<cftry>
<cfinvoke component="cfide.adminapi.runtime"
method="clearTrustedCache" />
<cfset msg = listAppend(msg, "Cleared Trusted Cache") />
<cfcatch type="any">
<cfset msg = listAppend(msg, "FAILED to clear Trusted
Cache") />
</cfcatch>
</cftry>
<!--- while onApplicationStart() is running, errors can be
thrown so move the beanfactory over during shutdown --->
<cfset bf = application.cs />
<cfset onApplicationStart() />
<cfset msg = listAppend(msg, "onApplicationStart(): Coldspring
reset") />
<!--- shutdown transfer cache to ensure memory is released --->
<cftry>
<cfset tf = bf.getBean('ormService') />
<cfset tf.shutdown() />
<cfset msg = listAppend(msg, "Transfer shutdown") />
<cfcatch type="any">
<cfset msg = listAppend(msg, "Unable to clear Transfer
cache") />
</cfcatch>
</cftry>
<!--- force individual Model-Glue reloads (suggestion from Sean
Corfield) --->
<cfset structDelete(application, "public", false) />
<cfset structDelete(application, "subapp1", false) />
<cfset structDelete(application, "subapp2", false) />
<cfset msg = listAppend(msg, "Model-Glue restarted") />
</cfif>
</cflock>
<cfif structKeyExists(application, "cs")>
<cfset ss = application.cs.getBean('sessionService') />
<cfset ss.getMessaging().addWarning(msg) />
</cfif>
</cfif>
With very little load (e.g., one user) I'm able to fairly consistently
reproduce AN error. Rarely the same error twice because the request is
hitting the reload at some random point in the cycle so it's always
different. I believe the renaming and shutdown of the Transfer cache AFTER
restarting Coldspring (and a new instance of Transfer) has cleaned up some
of my errors with ehCache reporting it has been shutdown.
My questions are:
* Is this the best way to reinit?
* Should the order be different? E.g., delete model-glue first, then
coldspring, then run onApplicationStart()?
* How do other people deploy Model-Glue apps and refresh their service
layer without interrupting end user sessions?
* The structDelete on the Model Glue framework was recommended by Sean
Corfield many moons ago; perhaps this is not the best way to trigger a
reload under 3.2 or for multiple sub apps?
Thanks!
Brian
--
Model-Glue Sites:
Home Page: http://www.model-glue.com
Documentation: http://docs.model-glue.com
Bug Tracker: http://bugs.model-glue.com
Blog: http://www.model-glue.com/blog
You received this message because you are subscribed to the Google
Groups "model-glue" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/model-glue?hl=en
