ColdFusion 10 update 3 released
Adobe ColdFusion 10 | Announcements | ColdFusion | Connector
[Update: The ColdFusion 10 update 3 will be pulled back shortly because of a few issues in the update. We are working on the fixing these issues. A new update will be released as soon as it is ready. Thank you for your patience and we apologize for the inconvenience caused]
We have released the ColdFusion 10 update 3 today. It fixes quite a number of important/critical issues, mostly related to connectors, in ColdFusion 10. This update can be downloaded and installed using the update installer mechanism in the administrator that we introduced in ColdFusion 10. Please make sure that you have applied the ColdFusion 10 mandatory update before applying this update. Also note that, you MUST re-configure your connectors after you have applied the update and restarted ColdFusion server.
Some of the important bugs fixed in this update are
- ColdFusion server is non-responsive when receiving a POST request with an XML body for a Web Service.
- Web sites fail with 'Service Not Available' error intermittently.
- Pages with multiple frames load the incorrect pages when updating frames, intermittently.
- Application pool on IIS 7.5 crashes due to certain Web Service requests.
- DateConvert("local2UTC", date) breaks other DateTime functions.
For more details on this release, please refer to this technote.
133 comments so far ↓
I have updated the coldfusion server update 3
and after that i am trying to reconfigure the connectors using wsconfig tool on my developer machine which is [Window 7 professional edition].But wsconfig UI is not coming .
I have to run it as administrator also but problem exists UI is not coming.
Should i have to re-start the system?
Thanks
Subodh
Does
C:\CF10\cfusion\bin>cfstart.bat
Error loading: C:/Program Files/Java/jdk1.6.0_31\jre\bin\server\jvm.dll
I have verified that my java path location is correct and that the jvm.config file is identical to the backup version.
Any thoughts on what might have gone wrong with my install? I have already run uninstaller.jar once (that brought me back to a working state). After I applied ColdFusion 10 update 3 again I continued to get error loading jvm.dll.
Thanks,
Bob
I have exactly your same problem.
I have not found any solution yet.
Cheers,
PG
Could you please chage the jvm.config to point to the JRE being shipped with ColdFusion and see if you still see the same issue? If that fixes it, it would mean that the java version you have is not correct and you would need to correct it.
Unfortunately I'm having the same problem with the CF JRE as well.
My JRE versions are as follows:
c:\CF10\dev10\bin>..\..\jre\bin\java -version
java version "1.6.0_29"
Java(TM) SE Runtime Environment (build 1.6.0_29-b11)
Java HotSpot(TM) 64-Bit Server VM (build 20.4-b02, mixed mode)
c:\CF10\dev10\bin>"c:\Program Files\Java\jdk1.6.0_31\jre\bin\java.exe" -version
java version "1.6.0_31"
Java(TM) SE Runtime Environment (build 1.6.0_31-b05)
Java HotSpot(TM) 64-Bit Server VM (build 20.6-b01, mixed mode)
Faulting application name: coldfusion.exe, version: 10.0.0.0, time stamp: 0x5052ed25
Faulting module name: ntdll.dll, version: 6.1.7601.17725, time stamp: 0x4ec4aa8e
Exception code: 0xc0000374
Fault offset: 0x00000000000c40f2
Faulting process id: 0x21a8
Faulting application start time: 0x01cdac1eaf2deb0b
Faulting application path: C:\ColdFusion10\cfusion\bin\coldfusion.exe
Faulting module path: C:\Windows\SYSTEM32\ntdll.dll
Report Id: f221b3a4-1811-11e2-ac98-0015f2d75c8f
Just mentioning FWIW. Again, CF10 Update 3 is running fine for me now.
Thanks,
-Aaron
What is the command that you used to install if you are installing from command prompt?
If you don't give full path of java.exe while installing from command prompt, your machine's default system path jre will be used to install.
In your system path 32-bit jre could be there but if your CF is 64-bit installation you have to use 64-bit jre explicitly to install the Hofix.
If you are applying from CF Administrator this happens automatically.
So, you if you invoke the Hotfix from command prompt with same jre as that CF is using as follows everything should be fine.
"c:\Program Files\Java\jdk1.6.0_31\jre\bin\java.exe" -jar hotfix_003.jar
Thanks,
Krishna
ColdFusion Team
What is the command that you used to install if you are installing from command prompt?
If you don't give full path of java.exe while installing from command prompt, your machine's default system path jre will be used to install.
In your system path 32-bit jre could be there but if your CF is 64-bit installation you have to use 64-bit jre explicitly to install the Hofix.
If you are applying from CF Administrator this happens automatically.
So, you if you invoke the Hotfix from command prompt with same jre as that CF is using as follows everything should be fine.
"c:\Program Files\Java\jdk1.6.0_31\jre\bin\java.exe" -jar hotfix_003.jar
Thanks,
Krishna
ColdFusion Team
I applied Update 3 via CF Administrator. I have more info now. The error I reported can be reproduced w/ these steps:
1) Stop CF via Windows Services applet
2) Start CF via Windows Services applet
Repeat steps 1 & 2 until Windows Services applet throws crash report saying coldfusion.exe crashed. Sometimes CF starts without error. Other times coldfusion.exe crashes during start attempt. It is random.
Thus, not installer-related. It is stop/start related.
Thanks!,
-Aaron
The error log for the update via administrators shows (errors only):
Installation: Unsuccessful.
109 Successes
0 Warnings
1 NonFatalErrors
1 FatalErrors
Action Notes:
Failed to copy hotfix files:C:\Users\xxx\480999.tmp\dist\cfusion: Failed to copy the hotfix files to the target location. Retry installation after ensuring that the server is not running or files are not locked by the server.
Install Log Detail:
Modify Text File - Single File: New File C:\Program Files\cfhotfix.txt
Status: ERROR
Additional Notes: ERROR - java.io.FileNotFoundException: C:\Program Files\cfhotfix.txt (Access is denied)
Failed to copy hotfix files:C:\Users\xxx\480999.tmp\dist\cfusion
Status: FATAL ERROR
Additional Notes: FATAL ERROR - Failed to copy the hotfix files to the target location:C:\ColdFusion10\cfusion
FATAL ERROR - C:\ColdFusion10\cfusion\bin\coldfusionsvc.exe (The process cannot access the file because it is being used by another process)
Thanks,
-Aaron
Installed Feature(s) Standard, J2EE of Adobe ColdFusion 10 Update 3
Summary
-------
Installation: Successful.
112 Successes
0 Warnings
0 NonFatalErrors
0 FatalErrors
Thanks,
-Aaron
1) Stopped all CF10 services via the Windows Services applet: ColdFusion 10 .NET Service, ColdFusion 10 Application Server, ColdFusion 10 Jetty Service, ColdFusion 10 ODBC Agent, and ColdFusion 10 ODBC Server
2) Once all the above services were stopped, then I launched Command Prompt as an Administrator and ran the following: C:\ColdFusion10\cfusion\bin\cfstart.bat
Error:
C:\ColdFusion10\cfusion\bin>cfstart.bat
#
# A fatal error has been detected by the Java Runtime Environment:
#
# EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x0000000076dd0895, pid=4816, t
id=2360
#
# JRE version: 6.0_29-b11
# Java VM: Java HotSpot(TM) 64-Bit Server VM (20.4-b02 mixed mode windows-amd64
compressed oops)
# Problematic frame:
# C [ntdll.dll+0x50895]
#
# An error report file with more information is saved as:
# C:\ColdFusion10\cfusion\bin\hs_err_pid4816.log
That error log is too verbose to paste into a blog comment.
I'm trying to create repro cases for some REST bugs I recently found, and this start/stop issue is now making that process take longer :)
Thanks,
-Aaron
1) Stopped CF via Windows Services applet
2) Waited a full 15 minutes
3) Attempted to start CF via Windows Services applet and got the same coldfusion.exe crash report
Thanks,
-Aaron
I have to create it manually once again .
This is really frustrating .....
Some issues:
1) The error says the page is named "Scheduler Tasks". Wrong - it's named "Scheduled Tasks".
2) The error says the Scheduled Tasks page is located under "Debugging & Logging". Wrong - it's been moved under "Server Settings" as of CF10.
3) I have ZERO scheduled tasks defined :) So what's this new error noise about?
"Error","pool-2-thread-1","10/17/12","03:46:09",,"Enable Cluster setup (ColdFusion administrator > Debugging & Logging > Scheduler Tasks). "
coldfusion.scheduling.CronServiceImpl$ClusterNotEnabledException: Enable Cluster setup (ColdFusion administrator > Debugging & Logging > Scheduler Tasks).
at coldfusion.scheduling.CronServiceImpl.getSchedulerInstance(CronServiceImpl.java:2210)
at coldfusion.scheduling.CronServiceImpl.pauseScheduler(CronServiceImpl.java:2277)
at coldfusion.scheduling.CronServiceImpl.updateTasks(CronServiceImpl.java:277)
at coldfusion.scheduling.CronServiceImpl.load(CronServiceImpl.java:238)
at coldfusion.server.ServiceBase.start(ServiceBase.java:59)
at coldfusion.server.CFService.setupCron(CFService.java:311)
at coldfusion.server.CFService.start(CFService.java:566)
at coldfusion.server.j2ee.CFStartUpServlet.startCFService(CFStartUpServlet.java:110)
at coldfusion.server.j2ee.CFStartUpServlet.init(CFStartUpServlet.java:86)
at javax.servlet.GenericServlet.init(GenericServlet.java:160)
at coldfusion.bootstrap.ClassloaderHelper.initServletClass(ClassloaderHelper.java:121)
at coldfusion.bootstrap.BootstrapServlet.init(BootstrapServlet.java:59)
at org.apache.catalina.core.StandardWrapper.initServlet(StandardWrapper.java:1266)
at org.apache.catalina.core.StandardWrapper.loadServlet(StandardWrapper.java:1185)
at org.apache.catalina.core.StandardWrapper.load(StandardWrapper.java:1080)
at org.apache.catalina.core.StandardContext.loadOnStartup(StandardContext.java:5001)
at org.apache.catalina.core.StandardContext.startInternal(StandardContext.java:5278)
at org.apache.catalina.util.LifecycleBase.start(LifecycleBase.java:150)
at org.apache.catalina.core.ContainerBase$StartChild.call(ContainerBase.java:1525)
at org.apache.catalina.core.ContainerBase$StartChild.call(ContainerBase.java:1515)
at java.util.concurrent.FutureTask$Sync.innerRun(FutureTask.java:303)
at java.util.concurrent.FutureTask.run(FutureTask.java:138)
at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)
Starting to lose track in this convo.
Tomorrow I can log the Update 3 breaks cfstart.bat 100% and prevents Windows Services applet from starting CF ~50% of the time. I'll prob do that when filing a few REST bugs.
We should probably test CF Updates more before releasing into the wild. Gotta go for now!
Thanks,
-Aaron
Thanks!,
-Aaron
It would be helpful if have a connect session to resolve the issue that you are facing.I have send a mail regarding same.
Thanks
-Evelin Varghese
www.eveysijo.com
For example:
We are accessing a page in:
D:/InetPub/Website1
And it says it can't find template from
D:/InetPub/Website23/includes/top.cfm
Now why is it even looking at website23 when we are on website1 URL?
Now to try to roll back the update... it seems Adobe might want to pull this update before more people are screwed.
Thankfully we're still CF8 in production:
Event Type: Information
Event Source: Application Popup
Event Category: None
Event ID: 26
Date: 10/17/2012
Time: 9:21:11 AM
User: N/A
Computer: xxx-xxxx
Description:
Application popup: Microsoft Visual C++ Debug Library : Debug Assertion Failed!
Program: C:\ColdFusion10\cfusion\bin\coldfusion.exe
File: f:\dd\vctools\crt_bld\self_64_amd64\crt\src\dbgheap.c
Line: 1317
Expression: _CrtIsValidHeapPointer(pUserData)
For information on how your program can cause an assertion
failure, see the Visual C++ documentation on asserts.
(Press Retry to debug the application)
For more information, see Help and Support Center at http://go.microsoft.com/fwlink/events.asp.
[Wed Oct 17 09:25:25 2012] [1656:2324] [info] jk_open_socket::jk_connect.c (626): connect to 127.0.0.1:8012 failed (errno=61)
[Wed Oct 17 09:25:25 2012] [1656:2324] [info] ajp_connect_to_endpoint::jk_ajp_common.c (1027): Failed opening socket to (127.0.0.1:8012) (errno=61)
[Wed Oct 17 09:25:25 2012] [1656:2324] [error] ajp_send_request::jk_ajp_common.c (1649): (cfusion) connecting to backend failed. Tomcat is probably not started or is listening on the wrong port (errno=61)
[Wed Oct 17 09:25:25 2012] [1656:2324] [info] ajp_service::jk_ajp_common.c (2629): (cfusion) sending request to tomcat failed (recoverable), because of error during request sending (attempt=1)
[Wed Oct 17 09:25:26 2012] [1656:2324] [info] jk_open_socket::jk_connect.c (626): connect to 127.0.0.1:8012 failed (errno=61)
[Wed Oct 17 09:25:26 2012] [1656:2324] [info] ajp_connect_to_endpoint::jk_ajp_common.c (1027): Failed opening socket to (127.0.0.1:8012) (errno=61)
[Wed Oct 17 09:25:26 2012] [1656:2324] [error] ajp_send_request::jk_ajp_common.c (1649): (cfusion) connecting to backend failed. Tomcat is probably not started or is listening on the wrong port (errno=61)
[Wed Oct 17 09:25:26 2012] [1656:2324] [info] ajp_service::jk_ajp_common.c (2629): (cfusion) sending request to tomcat failed (recoverable), because of error during request sending (attempt=2)
[Wed Oct 17 09:25:26 2012] [1656:2324] [error] ajp_service::jk_ajp_common.c (2648): (cfusion) connecting to tomcat failed.
[Wed Oct 17 09:25:26 2012] [1656:2324] [info] jk_handler::mod_jk.c (2725): Service error=-3 for worker=cfusion
[Wed Oct 17 09:25:32 2012] [1656:2324] [info] jk_open_socket::jk_connect.c (626): connect to 127.0.0.1:8012 failed (errno=61)
[Wed Oct 17 09:25:32 2012] [1656:2324] [info] ajp_connect_to_endpoint::jk_ajp_common.c (1027): Failed opening socket to (127.0.0.1:8012) (errno=61)
[Wed Oct 17 09:25:32 2012] [1656:2324] [error] ajp_send_request::jk_ajp_common.c (1649): (cfusion) connecting to backend failed. Tomcat is probably not started or is listening on the wrong port (errno=61)
[Wed Oct 17 09:25:32 2012] [1656:2324] [info] ajp_service::jk_ajp_common.c (2629): (cfusion) sending request to tomcat failed (recoverable), because of error during request sending (attempt=1)
[Wed Oct 17 09:25:33 2012] [1656:2324] [info] jk_open_socket::jk_connect.c (626): connect to 127.0.0.1:8012 failed (errno=61)
[Wed Oct 17 09:25:33 2012] [1656:2324] [info] ajp_connect_to_endpoint::jk_ajp_common.c (1027): Failed opening socket to (127.0.0.1:8012) (errno=61)
[Wed Oct 17 09:25:33 2012] [1656:2324] [error] ajp_send_request::jk_ajp_common.c (1649): (cfusion) connecting to backend failed. Tomcat is probably not started or is listening on the wrong port (errno=61)
[Wed Oct 17 09:25:33 2012] [1656:2324] [info] ajp_service::jk_ajp_common.c (2629): (cfusion) sending request to tomcat failed (recoverable), because of error during request sending (attempt=2)
[Wed Oct 17 09:25:33 2012] [1656:2324] [error] ajp_service::jk_ajp_common.c (2648): (cfusion) connecting to tomcat failed.
[Wed Oct 17 09:25:33 2012] [1656:2324] [info] jk_handler::mod_jk.c (2725): Service error=-3 for worker=cfusion
Running the uninstaller did... absolutely nothing.
Manually uninstalling as per the directions (sort of... like all the old ColdFusion updates pre-CF10 you have to do what it means and not what it says... i.e. don't overwrite folders... copy the contents of the folders into the other folders) actually worked and I'm back to Update 2 and a functioning site.
Updated the connector and now got segmentation fault (11) in the apache error.log.
Replaced wsconfig.jar with the one from the hf3-backup-dir and reinstalled connector - everything working again...
Seems like somethings wrong here...
Oct 16, 2012 22:02:05 PM Information [pool-2-thread-1] - C:\ColdFusion10\cfusion\logs\scheduler.log initialized
Oct 16, 2012 22:02:05 PM Error [pool-2-thread-1] - Enable Cluster setup (ColdFusion administrator > Debugging & Logging > Scheduler Tasks).
Oct 16, 2012 22:02:05 PM Error [pool-2-thread-1] - Enable Cluster setup (ColdFusion administrator > Debugging & Logging > Scheduler Tasks).
These appear each time I start CF, so it would seem it is "initializing" the scheduled tasks on each start (not good). Also, I am not using clustering (nor may Aaron be), so the reference to that seems odd.
But back to some of the earlier comments, I'm not experiencing any of those other problems (the UI starting, CF not starting, nor errors trying to stop and start repetitively). Not denying or deprecating those who are experiencing those problems, just offering at least one voice noting that they're not a universal experience.
What we can see for sure is that the application name variables are occasionally containing application names from other sites on that server.
This is Standard version.
And yes, all scheduled tasks were deleted.
So far no issues starting CF in general though.
But random inclusions from other sites is bad enough.
I also had a lot of odd collisions as mentioned by tohers above, where my virtual hosts were mixing up during sessions. Someone would visit a site, but it would call cgi.server_name of some other vertual site, or the error debugging would show paths of both the correct site and some other virtual host on the server.
I removed all connections via wsconfig to IIS, then readded ALL and restarted during the whole process and it now seems OK though I am not confident that the issue is corrected.
@Charlie, Thanks for confirming the start-up failure isn't universal. I agree CF should _never_ run tasks upon CF start or .car deploy. I wonder if that is related to this:
----
Paused tasks having onmisfire="fire_now" will run on every CF start (#3141655 - marked ToTest/Fixed) and .car deploy (#3178809 - marked Unverified).
----
Basically, tasks have always misfired on every CF start and .car deploy. Which means that if they use onmisfire="fire_now", paused tasks will actually run at those times.
Logged those in March and April.
Charlie, would you mind also verifying if scheduled tasks are deleted on CF start? Just curious if that is/isn't universal.
Thanks!,
-Aaron
I am experiencing CF startup problems, deleted scheduled tasks, all 83 of them, random includes in the wrong site like Jason L reported.
Without the update CF not stable with constant "Service Not Available" errors. With the update constant startup errors, errors due to templates from wrong site being pulled in and no scheduled tasks. Either way, not usable.
Regarding other issues, we have not been able to reproduce them so far. We are still looking into them and we might get in touch with some of you separately to get more details.
Meanwhile, could you please confirm if you had RECONFIGURED the connectors after applying the update? Since this update is primarily for connectors, it is absolutely CRITICAL that you reconfigure it. Without that, server would get in an inconsistent state and can result in some unpredictable behavior.
In other words, "No, I personally did not reconfigure the connector as I assumed the installer would leave the server in a functioning state."
@All, Could you please reconfigure the connectors after restarting the server and let us know if you still see the issues?
All issues I reported continue unchanged after unconfigure/reconfigure via wsconfig.exe. I did not notice this reconfigure (2nd bullet on technote) requirement b/c this requirement was not mentioned in CF Admin nor in the Updater UI. Can requirements like this please be mentioned in the CF Admin and on last screen of Updater UI? Basically, I understand it's my fault for not scrolling down on the technote page to see that. However, I feel CF Admin and Updater UI could also help out here.
Thanks,
-Aaron
I've filed ER #3348487 (Updaters should always create a .car file).
If Updaters always created a .car file, then users of Update 3 would've been more easily able to restore back their scheduled tasks.
I understand users can create this .car file manually (and via special workaround in Standard Edition) but I feel this should be done automatically.
Thanks,
-Aaron
I've confirmed it is safe to unconfigure/reconfigure via wsconfig.exe in IIS7+ (the applicationHost.config file is identical before/after the unconfigure/reconfigure). It appears the connector changes were made to isapi_redirect.dll and changes its date modified from August 9, 2012 to September 26, 2012.
Again, just confirming it is safe and doesn't alter site configurations.
Thanks,
-Aaron
Installation summary (ColdFusion won't start at all):
Windows 2003 R2 SP2 64bit (same as before the upgrade)
ColdFusion 64bit (Enterprise)
Java 64bit (jdk1.6.0_35: same as before the upgrade)
Apache 2.2.22 (32bit: same as before the upgrade)
Everything works fine with update 2. Update 3 makes it completely non-functional.
Installation summary (ColdFusion works fine):
Mac OS 10.7
ColdFusion Developer Edition
Latest Apple supplied JRE
Latest Apple supplied JRE
I did NOT reconfigure the connectors on this Mac and it works fine (or at least it starts. I have not checked for the application errors other have reported).
Thanks,
-Aaron
Service Temporary Unavailable
Jakarta/ISAPI/isapi_redirector/1.2.32 ()
issue pop up again. We need to rollback to cf 9.1, which is still installed on the server in question, just disabled. Are there any directions on performing a graceful rollback?
And if you did indeed to a "remove" of all and "added them back", then you would not really know from doing that if it "did or did not work", from anything you could easily look at. So please consider doing that and see if it has any impact.
If it still is happening, here is an "out of the box" suggestion not related to the update: in the CF Admin "Caching" page, do you have "cache web server paths" checked? That's not touched by the update (or certainly shouldn't be), but some of what you describe is what happens when that's checked and you have multiple web sites (which is why it says under that setting those running multiple web sites should not turn it on.)
I appreciate that neither of these may help you. Just offering them in case they may.
We checked, unchecked and then cleared all caches and still have the same issues. But I do agree with you Charlie it's very much like the paths are cached. Do you know if there is an xml setting or some place where that entry would be saved that i could look at it?
And for that cache web server paths setting, it's in neo-runtime.xml as <var name='cacheWebServerPath'><boolean value='false'/></var>. (In case this blog does not render html/xml for display, that was the xml element in the file named cacheWebServerPath, which as a boolean true or false value (and should be false, to equate to not checked in the interface.)
Running uninstaller now for this update.
I managed to install this hotfix correctly today. When I ran the server update from my "CFIDE" instance (the instance that contains the "Enterprise Manager" tab in CFAdmin) the process proceeded correctly. When I ran the auto-update process from my "real" instance (the instance using actual code) the update process failed and I resorted to running the jar manually.
I have also confirmed that when I ran the update manually yesterday I was using the correct java versions; I'm not really sure what was causing my problems.
Second, the fact that the autoinstaller works only from within the Admin of the cfusion instance is discussed in the hotfix technote. It says that you can either run it there and choose the other instances in the checkbox offered (in the update installer) or, as you did, run it manually from the command line. (See the note starting, "The update can only be installed from the Administrator of cfusion instance. To apply the update to individual instances...")
Hope that's helpful.
For everyone who reported issue on coldfusion.exe / window services, please verify the following steps at your machine
1. Make sure you have used 64 bit java for installing the hotfix. If you have done that cross check the size of coldfusion.exe should be 533kb and coldfusionsvc.exe should be 132kb.
2. Use 64 bit dependencywalker (download from http://www.dependencywalker.com/) and drag and drop the coldfusion.exe. Check dependency walker is loading the coldfusion.exe without error. (Ignore the warnings)
3. Make sure you launch coldfusion.exe as Administrator / in AdminGroup / Run as Administrator. As coldfusion.exe internally needs to communicate to window service manager, it is important you have sufficient permission when your launching the coldfusion.exe
4. Make sure jvm.config pointing to the 64 bit JVM .
5. Please restart the machine once which could solve the ntdll.dll issue, which people are facing when running the coldfusion.exe
Thanks
Kiran Sakhare
Please provide details of your configuration
CF version - 32/64 bit
IIS /Apache - if Apache 32/64 bit?
Platrform - Widows/Linux/Mac
Thanks
Evelin
thanks
-Evelin
Not running on the latest connectors causes all sorts of drama.
@Adobe On that note, you may want to check the tool. "Remove" and "add" did nothing for me - remove didnt remove and add didnt add. Also, surely it would be better for the "Web Server Configuration Tool" to have an "update" button as apposed to having to "remove" and then re-"add" the connectors. Reason being, there may be a bunch of sites configured in IIS and the only option for adding the connectors, is one-by-one or to all. Possibly one might not want all sites configured for CF and if there are lots of them, configuring one-by-one is not an easy option.
As for your challenge with the add/remove doing "nothing", this is a common experience for those on Windows 7 or 2008, if you fail to notice that you must run the tool with "run as administrator". It's not enough to "be" an administrator. When you click Start>Programs>Adobe>ColdFusion 10, you then need to right-click on Web Server Configuration Tool and choose "run as administrator".
Sadly, if you don't, the interface will seem to otherwise "work", but as you've experienced, it won't make the needed changes.
Besides your helpful suggestion to Adobe about a possible "edit" button, it would also be nice if the tool could check if it is running as Admin. (The Mandatory Hotfix was made to do this, so these Java applets clearly can do such a check.)
@Jason L, I haven't yet tried simulating the issue where CF tries to load templates from other sites.
If you have recieved my mail please reply back.
We are looking into these issues.
-Evelin
and run Remove_ALL_connectors.bat as Administrator
and then IIS_connector.bat as administrator as well and that should fix all problems?
I guess its good to know what broke with the update, but its also good to know what didnt break.
My server is windows 2003 server with iis 6.
It is an intranet server with 30+ multi-homed sites.
All sites work correctly and did nothing more than apply the update thru the CF administrator.
My exact step were;
- Apply mandatory update - which failed because of Java version
- updated Java version to 6.(latest)
- reapply mandatory update - successful
- in CF Admin, clicked download and install
- update completed.
- rebooted server (just because)
- test successful installation.
After applying the update, reconfigure the connectors using wsconfig tool. It is in{cf_install_home}/cfusion/runtime/bin.
NOTE we rolled back the update because of all the issues we had.
If I start coldfusion from a command line it works fine.
C:\ColdFusion10\cfusion\bin\coldfusion.exe start
Starting from the service fails.
I did not try starting it using command line etc. I know that when I rebooted it, everything functioned correctly.
@Ryan - I did not reconfigure the connectors and because its functioning as it should, I'm not going to. (if it aint broke, dont fix it)
@Vance - I believe that all updates are supposed to be initiated thru CF Admin page, left hand side, Server Update, click UPDATEs. Use 'Available Updates' tab - Follow from there.
I don't believe it does anything until you rerun the connector. Look at modification dates in the following location
C:\ColdFusion10\config\wsconfig\1
1) Last night, I loaded 5 different locally-hosted (local CF10 Update 3 dev server) sites in different tabs. (tab1=www2.site1.com, tab2=www2.site2.com, .., tab5=www2.site5.com) (note: www'2' is local dev).
2) Last night, on tab 1 I navigated about site1 and saw no issues. Then switched to tab 2 and navigated about site2 and saw no problems. Continued same for the remaining 3 tabs and saw no problems.
3) Left the tabs sit open over night.
4) Just now I thought I'd try something different: I quickly jumped thru each tab and hit F5 on each (to cause all 5 sites to load at the same time). Guess what happened haha..
5) Tabs 1-3 loaded with no problems (www2.site1.com, www2.site2.com, and www2.site3.com loaded properly).
6) Tab 4 loaded www.site3.com (the LIVE version of tab 3's site!) w/ no error. (BAD! b/c it should've reloaded www2.site4.com)
7) Tab 5 loaded www.site3.com (again, the LIVE version of tab 3's site) w/ CF exception "Could not find the included template foo/bar.cfm." and the exception message also said "Referrer http://www2.site5.com/some/path/here"; (BAD! b/c it should've reloaded www2.site5.com) (and that Referrer really boggles my mind - CF is redirecting the request!?)
OK, I'm looking thru log files right now. I'll follow-up.
Thanks,
-Aaron
Just to be clear, no caching enabled. Still looking thru logs.
CF Admin does not have any caching enabled.
Install Steps as Follows:
1. Installed in CF Admin
2. Run the wsconfig tool as Administrator
3. Rebooted Server (just in case)
This isn't a permission problem (good thought though). Confirmed by having the service run as the same user that I can launch coldfusion.exe via the command line. The service still fails to launch when running as that user or local system. It's just plain broke.
And yes, I could see how it might be that that would influence how things would work for you one way when starting from the command prompt and another way from services, even when both as the "same user". Please do just humor us, to know if this is or is not an issue.
Similarly, @bobstamp, you say that "I did not reconfigure the connectors and because its functioning as it should, I'm not going to. (if it aint broke, dont fix it)" Again, looking over your comments, I'm trying to understand: are you saying (based on your first message) that you have no problems at all after running the updater? I'm just trying to discern if this subsequent conversation is then related to some thinking you did have a problem. :-)
Finally, @jason, in that you are still having problems, again I would ask: when you re-configured the web connector, did you "run it as admin"? I had asked this back in #53 but I don't see you responding to it, so again just want to make sure.
I'd like to clarify also, in case anyone may feel like I'm "sounding as a broken record", it's just that when these sort of things blow up, and there's a whole lot of comment from many people, it can be easy for key points to fall through the cracks. Especially when some people have problems that others do not (even if "a few others do also"), it just seems wise to keep pressing to make sure we're all on the same page.
Doing so not only helps (hopefully) the one being spoken to, and others who may have the same problem, but it may also help Adobe and others who are trying to assist. :-)
(I was torn about just waiting to see if you did respond, but I didn't want to forget and have you miss them.)
So just a heads up, if the thoughts may help.
@Paul Karlin, I'm doing this testing on CF10 Update 3
@Jason L, I'm testing on a similar setup: x64 Windows 7 SP1 with IIS 7.5. Thanks for noting that it happens even w/ navigating about a site.
@Charlie, I don't have clustering or load balancing anywhere
Thanks again!
We eagerly await Update 4.
But with so many (now over 90), it's not a bad idea to offer that simple clarification, so thanks. It seems so far that that's the only one that Adobe has confirmed is happening to "everyone".
But I see they have now added some "update" text to the opening of the blog entry. The opening sentence now reads, "the ColdFusion 10 update 3 will be pulled back shortly because of a few issues in the update."
It may be more helpful if they'd specifically mention at least this one known issue, if not a little more on what other "few issues" are, if they are now confirmed or highly suspected by Adobe, as they work on things.
Until then, let's help at least a cautious reader who sees that updated text might be compelled to read through the comments.
Along those lines, I do think it may be helpful if Adobe would bold that updated text now at the top. Some people tend to glaze over text and just click any links they see (especially as here where there it the link offered at the end of the entry to the technote for "more details").
Finally, speaking of that technote, I see that currently there is no mention at all (that I could see) of these "known issues" and that it would be "pulled back".
It would really seem appropriate that an update to that be made there (and made prominent as well).
Hope that's helpful.
"...if not a little more on what *the* other "few issues" are..."
and
"Until then, let's *hope* at least a cautious reader who sees that updated text..."
Also, if Adobe folks may not have read that whole last comment, I offered suggestions about the updated text made on this entry, and the lack of any update on the updater's technote.
FWIW, I work as an independent consultant, helping people with CF server troubleshooting. And the problems that these folks have are not CF10-specific, but are generally just the same sort of things I help folks with every day, on various versions of CF.
And I'll add, lest anyone take that to quip, "wow, there are people having so many problems with CF that Charlie has customers every day?", the fact is that in many instances, the problems end up NOT being CF at all.
I help people use available diagnostics, whether logs, tools, or otherwise, to first understand the nature of the problem, then identify the root cause, then apply a solution, and wait to see how that goes.
I'm just saying that often the problems are not "inherently issues with CF itself". They are often issues due to configuration (whether of CF or other things), or sometimes due to code (but not as often as most might think), or mostly due to unexpected load stressing things and never being noticed (whether spiders, bots, hackers, monitoring tools, pci compliance scanners, and so on.)
But yes, it may be interesting to know who here is running it in prod. It may not really be representative of all CF10 prod customers (as I can tell you that many people just never bother to do updates, so may be happily running it and not yet tried any of the updates.) But still, an interesting question nonetheless.
But yes, since I do fill in that request here in the comments for "website", that makes my name be a link on all my comments. And on that site, www.carehart.org, I do offer contact info. But to save you or anyone the trouble of looking, since you mentioned it (and may have been asking), my address is charlie (at) carehart.org.
I hope I could help them, and I appreciate the referral. Have them be sure to let me know that you passed them on to me. :-)
I did verify that the scheduled job are removed upon ColdFusion Application Service restart.
All my sites on the server are functional. As i said before, I installed as stated in comment 67 above. The pages report 'ColdFusion Server Standard 10,0,3,283145' and all of the sites 'appear' to be functioning correctly.
The isapi_redirect.properties file contains:
--------------------------------------
extension_uri= /jakarta/isapi_redirect.dll
log_file= C:\ColdFusion10\config\wsconfig\1\isapi_redirect.log
log_level= info
worker_file= C:\ColdFusion10\config\wsconfig\1\workers.properties
worker_mount_file= C:\ColdFusion10\config\wsconfig\1\uriworkermap.properties
iis_buffer_enable= true
--------------------------------------
And the isapi_redirect.log contain: (snippit)
------------------------------------------
[Thu Oct 18 23:31:19.026 2012] [4768:4248] [info] init_jk::jk_isapi_plugin.c (2792): Jakarta/ISAPI/isapi_redirector/1.2.32 () initialized
------------------------------------------
Now obviously, i dont know whether 1.2.32 is the correct redirector for this update. So if there are issues, i don't want to run the config tool as it may make changes that cause it to fail.
One note.... when i upgraded from CF9 to cf10, it was a frightful experience. I was on the phone for many hours with support where it would NOT install correctly. Unlike its predicessor cf7 to cf8, cf8 to cf9, all were 'click and done' upgrades. The only way i could get CF10 to work correctly was to uninstall all lingering versions of cf8 and cf9 and do a 'out of the box' clean install. But since i've done that, we haven't experienced many problems.
I've been doing this for a while as I'm the proud owner of Allaire Version 1.5 on 3 1/2" disks and have installed every version in-between.
Sent 10/19/2012 00:04 EDT
I was able to reproduce the issue where CF mixes up simultaneous requests for identical file paths in different sites. I've logged the repro case as #3349683 and Adobe knows about it and is working on it. This issue only affected Update 3 (Update 2 did not have this issue).
I was also able to reproduce to Adobe the issue where CF startup sometimes fails. Adobe is working on this as well.
And, of course, the scheduled tasks issue is also known and a fix has been verified for that.
Are there any other issues?
Many thanks to the CF Team for working on these issues! And ColdFusion Rocks!!!!
Thanks!,
-Aaron
That's on Linux64 with apache 2.2.
Don't know if Update3 should have fixed this, as the connector totally failed.
I'm back on hf2 now...
I also had lost my dozen scheduled tasks after installing update 3. I've noticed that lib/neo-cron.xml still had the scheduled task data, but nothing appears in the Administrator. Is there a way to recover the tasks from the xml file? Or is a better option to recreate them all? Thanks,
We have been able to identify and fix these three issues in the update 3.
- scheduled task not showing up in the admin
- ColdFusion windows service crashing/ColdFusion not starting up in some cases
- Mixing up of paths for simultaneous requests for identical paths in multiple sites.
A big thanks to all of you specially Aaron, Ryan and Sean for helping us investigate the issues.
There is one more issue regarding Directory index not working on apache which some of you have reported. Unfortunately we are still not able to reproduce it here locally and we would need your help for that. Those who are seeing this issue, can you send your httpd.conf and mod_jk.conf to 'rukumar' <at> adobe before and after applying the update 3?
Just to make sure, we have covered everything, are there any more issues?
I believe the only other two issues others mentioned were:
1st issue
---------
Joerg Zimmer mentioned a urlSessionFormat()/jsessionID issue that was introduced in CF10 Final and there is a forum discussion going here: http://forums.adobe.com/message/4780419
I'm unfamiliar w/ Apache, so hopefully others are able to help (I see Charlie and Steven have been helping already - very cool). Since that issue was not introduced in Update 3, I'd say proceed w/ Update 4 (since U4 is just a fixed re-post of U3) and perhaps Adobe can offer Joerg some assistance in that thread (it appears to be a backward-compat issue introduced in CF10).
2nd issue
---------
I see that some have mentioned having difficulty w/ wsconfig.exe. I haven't noticed any issue. Hopefully others can post further about the details of this.
Basically, I'm not sure what else to test. And I have a pretty busy weekend (child's b-day :))
And thanks Rupesh and the rest of the team and everyone here for working together!!
Thanks,
-Aaron
Currently there is a Microsoft IIS7+ appcmd.exe limitation which prevents virtual directories from being inherited into newly defined sites, even when choosing "All sites" when running wsconfig.exe. So adding the /CFIDE & /jakarta virtual directories will still need to be done manually (even after having previously chosen "All sites"). All the other settings tho should be inherited.
If there is any IIS7+ wsconfig.exe issue in migrating from CF9=>CF10, it is likely due to some CF9 settings still existing in the web.config file in the site's root. If that is the case, the CF-related settings can be removed from the web.config file. (Thankfully, CF10 doesn't use web.config anymore) If the web.config file contains only CF-related settings, then it can simply be deleted.
I used to use my own batch script instead of wsconfig.exe (b/c I wanted to configure IIS7+ for CF9 purely at server-level aka applicationHost.config). Now that CF10 does this I've ditched my batch script and use wsconfig.exe again. Thanks Adobe!!!! :)
HTH!,
-Aaron
I knew it was working fine a couple of weeks ago. So, I uninstalled the update and the problem was gone.
Hope the CF developing team would not miss to include this in their check list for the upcoming update.
I've confirmed that Update 3 breaks cfcompile -deploy.
Worked in Update 2 so should be fixed in Update 4.
Thanks,
-Aaron
Logged as #3350377 w/ simple repro.
https://bugbase.adobe.com/index.cfm?event=bug&id=3350377
Thanks!,
-Aaron
"After applying the update, reconfigure the connectors using wsconfig tool. It is in{cf_install_home}/cfusion/runtime/bin."
Nuh-uh. Given this is a repetitive taks (ie: ripe for computer automation), why is this a manual human-actioned task? The whole raison d'etre of this automated update system is to remove the need for a human to do this stuff.
What's the story here?
--
Adam
Thanks for the confirmation(testing) and report(bugbase) regarding the cfcompile issue.
You're exactly right, for even more reasons :)
1) The only wsconfig.exe command that needs ran post-Updater is the "-upgrade" command. However, if human runs the wsconfig.exe tool, the human will be unable to run the -upgrade option b/c..
2) The wsconfig.exe tool lacks an "Upgrade" button. It has these buttons: [Add...] [Remove] [Exit]. It should have these buttons: [Add...] [Upgrade] [Remove] [Exit]. Additionally, if human does do the Remove/Add method of reconfig, then permanent changes are made to the uriworkermap.properties file (and possibly the web server's config file - which, for IIS7+ is applicationHost.config) b/c..
3) The wsconfig.exe tool does not create a backup of the {CFInstallRoot}\config\wsconfig and the web server's config file (ex: applicationHost.config for IIS7+). Therefore, user is unable to restore back the applicationHost.config as well as loses any uriworkermap.properties customizations, such as..
4) Currently CF has case-sensitivity issues. One of these issues is worked-around via customizations to the uriworkermap.properties file (in addition to web.xml). So, human use of wsconfig.exe to add/remove the connector will restore one of CF's case-sensitivity issues. Additionally, uriworkermap.properties is used when user wants sites to support custom SES URLs (in addition to the SES URLs that CF supports out-of-the-box). So those would be lost as well.
5) And user may not even know a post-update connector reconfig is necessary, b/c the last screen of the Updater UI does not mention the need for a connector reconfig. (it's mentioned on the technote, but could be missed - I'm guilty of this =P)
Therefore, the simple fix is: Updaters should automatically run "wsconfig.exe -upgrade -v" command. No user action needed.
(I'll log a bug for the missing [Upgrade] button in the wsconfig.exe UI.)
Thanks,
-Aaron
-Aaron
Well that's interesting to know. So Rupesh (or some equivalent Adobe bod)... what's the story here? Why is this step not automated?
--
Adam
Note added says:
Users did not know that creating new tasks would cause their previously-defined tasks to be deleted from neo-cron.xml - b/c users did not know that their previously-defined tasks still existed in neo-cron.xml (b/c Updater 3 hid the tasks in CF Admin).
Worse, Standard Edition still lacks the .car creation/deploy feature. I thought there were public rumors that this would be fixed in CF10, but that didn't happen. So these customers likely did not have a .car file containing their scheduled tasks. If they'd already created new tasks (b/c Updater 3 hid them in CF Admin), then these customers experienced data loss as a result of an Updater. That is NOT GOOD!
It is *very important* that Updaters ALWAYS create a .car file of all CF Admin settings at the start of the update process.
It is also important that Standard Edition include the "Packaging & Deployment > ColdFusion Archives" option in CF Admin starting w/ CF11. Customers who purchased Standard Edition should never again have to suffer due to data loss caused by an Updater.
--
Adam
@adam - also voted for #3350535
@Adam, voted for #3350535
--
Adam
foo = now();
bar = dateConvert("local2UTC", now());
In CF9, <cfquery> sends #bar# to db as the converted value dateConvert("local2UTC", now())
In CF10, <cfquery> sends #bar# to db as the original value now()
Update 3 adds a JVM arg to regain CF9 behavior instead of restoring CF9 behavior.
IMO, user would expect the UTC timestamp to be persisted. Maybe I'm missing something. Thoughts?
So currently no way to effect CF9 behavior (honoring of timezone).
mod_jk.log
[Mon Oct 29 10:16:11 2012] [4962:3086186256] [info] init_jk::mod_jk.c (3322): mo
d_jk/1.2.32 () initialized
[Mon Oct 29 10:16:31 2012] [4962:3086186256] [info] init_jk::mod_jk.c (3322): mo
d_jk/1.2.32 () initialized
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [error] jk_open_socket::jk_connect.
c (448): socket() failed (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [info] ajp_connect_to_endpoint::jk_
ajp_common.c (1047): Failed opening socket to (127.0.0.1:8012) (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [error] ajp_send_request::jk_ajp_co
mmon.c (1669): (cfusion) connecting to backend failed. Tomcat is probably not st
arted or is listening on the wrong port (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [info] ajp_service::jk_ajp_common.c
(2684): (cfusion) sending request to tomcat failed (recoverable), because of er
ror during request sending (attempt=1)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [error] jk_open_socket::jk_connect.
c (448): socket() failed (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [info] ajp_connect_to_endpoint::jk_
ajp_common.c (1047): Failed opening socket to (127.0.0.1:8012) (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [error] ajp_send_request::jk_ajp_co
mmon.c (1669): (cfusion) connecting to backend failed. Tomcat is probably not st
arted or is listening on the wrong port (errno=22)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [info] ajp_service::jk_ajp_common.c
(2684): (cfusion) sending request to tomcat failed (recoverable), because of er
ror during request sending (attempt=2)
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [error] ajp_service::jk_ajp_common.
c (2703): (cfusion) connecting to tomcat failed.
[Mon Oct 29 10:16:41 2012] [5234:3086186256] [info] jk_handler::mod_jk.c (2748):
--
Adam
--
Adam
All the issues that were reported with update 3 have been resolved. Thanks a lot to each one of you for your feedback and your assistance in helping us identify and resolve the issues.
Many of you have also helped us in verifying the early version of the update 4 and we really value and appreciate your help for that. You would need to un-install that pre-release version of update and then re-download/install the final version of it.
Thank you for your work on update 4!
I just wanted to report an issue I had while installing:
I uninstalled my earybird hf4 update and changed the update URL to default.
Installing hf4 went smooth, but CF-Admin kept informing me about a new update. I saw that hf4 was listed in the update-level, but it has not been unpacked in the hf-updates folder. Manually install hf4 via "java -jar hotfix_004.jar" solved that - but this should not be the way to go...
Also my issue with the appended ;jsessionid is not yet solved - but my bug report is still open, so I'll have to wait for that...
Joerg
I outlined the steps here: thefusionator.com/index.cfm?blogID=4
-Michael
Leave a Comment