Status/Resolution/Reason: Closed/Fixed/
Reporter/Name(from Bugbase): Jake Hand / Jake Hand (Jake Hand)
Created: 05/08/2013
Components: Installation/Config, Connector
Versions: 10.0
Failure Type:
Found In Build/Fixed In Build: Final /
Priority/Frequency: Major / Most users will encounter
Locale/System: English / Win 2008 Server R2 64 bit
Vote Count: 2
Problem Description: In a multi-tenant Windows Server 2008/IIS 7.5 environment (with separate application pools for each site), ColdFusion 10 intermittently displays the following error:
Service Temporary Unavailable!
The server is temporarily unable to service your request due to maintenance downtime or capacity problems. Please try again later.
Jakarta/ISAPI/isapi_redirector/1.2.32 ()
Steps to Reproduce: This intermittent issue affects some sites on a server, and when it occurs you only have to try browsing to the affected site to experience the issue. It will take forever for the site to load, and it will eventually just show the error above.
Actual Result: Instead of the site loading properly, the Jakarta error above is displayed.
Expected Result: The site's page should load without error.
Any Workarounds:
When the error occurs the only solution to get all sites working again is to reset CF and IIS.
Because the CF team mentioned that the IIS-CF connector needs to be tuned based on the server's load, we have spent a lot of time tuning the connector settings for our environment. First we tried increasing the "max_reuse_connections" attribute for the connector in its workers.properties file. This would delay the occurrence of the error, but it only provided temporary relief.
Then we tried enabling the "connection_pool_timeout" setting as recommended by the Adobe ColdFusion team's blog. This too would delay the error, but no matter what type of timeout was used the error would eventually come back, as if threads were not being timed-out by the connector. We've tried timeouts short (as low as 30 seconds) and long (as long as the 10 minutes recommended by Tomcat's Timeouts How-To guide) with no real success. Eventually the error always appears. When the error occurs on a site our JVM heap is in great shape and our performance monitor (Seefusion) shows no hung requests queuing up.
Lastly, we've tried also setting the "connection_pool_size" and the "connection_pool_size" already provided by Tomcat's web server connector. These too seem to provide temporary relief, but no combination of any of the above settings seems to keep CF stable long-term. Additionally, because there is almost no documentation on the "max_reuse_connections" attribute (which seems specific to CF 10) it is not evident how that setting relates to the "connection_pool_size" setting.
----------------------------- Additional Watson Details -----------------------------
Watson Bug ID: 3556864
External Customer Info:
External Company:
External Customer Name: jakefusion
External Customer Email:
External Test Config: My Hardware and Environment details:
- Windows Server 2008 64-bit
- IIS 7.5
- ColdFusion Enterprise 64-bit "Standalone" installationAttachments:
Comments: