Miscellaneous errors and comments on SP3 (re-released version)
For those who are considering SP3 overall I think its a great release but here are a few things I learned in the process.
* It did not install correctly the first time. After a successful installation the nodes would not start up due to a bean loading exception. I had to copy one of the three nodes, replace the other two node folders (we have 3 total under <PPM_HOME>\servers) then reinstall. I thought this was a fluke on our dev box, but it did it on production as well. Perhaps there was a difference within the folder contents of the nodes that I was unaware of.
* The error 'dirkf' pointed out with the single quotes in the field names generated exceptions that prevented creation of requests. We've always had single quotes in one of our fields so this was a bit surprising. Glad he knew the answer!
* If you have a large staffing profile and the 'Compare to Work Plan' button errors out with the "more than 1000 lines" oracle error - that is not fixed in SP3. It is still pending under QCCR1L39510.
* The logs are much more detailed and the start process does a better job validating the server.config file. If you have any duplicates in your config like we did, the app may not start until you fix those errors. The logs will tell you where the duplicates are.
* Logs also roll correctly on windows - this was mentioned in the fixes. I believe it clears the logs on startup though so I have to remember to back them up if I need a record of why the node shut down.
* Application nodes start/stop much faster.
* You may need to reapply some patches, but the hot deploy of JSP files does work now if you use that option.
* We still have memory use issues on the user node (does not seem to do garbage collection/cleanup properly on our primary user node - could be windows issue), but the services node is running with much less memory use which is great.
I'd be curious to hear feedback from others and what sort of good/bad things you may be seeing.
We have a few errors in our logs that I'm curious as to what they mean. Perhaps some of you here have seen these as well. They do not seem to cause a problem, but they generate a lot of log noise.
Services node errors (new since SP3).... ERROR :ppmHeavyServiceListenerContainer-1:com.mercury.itg.pm.cost.util.ProjectCostUtil:2011/07/03-09:33:04.936 CDT: No TM effort contour found for TM work unit: 1202033 ERROR :ppmHeavyServiceListenerContainer-1:com.mercury.itg.pm.cost.util.ProjectCostUtil:2011/07/03-09:33:05.061 CDT: No TM effort contour found for TM work unit: 1238004 ERROR :ppmHeavyServiceListenerContainer-1:com.mercury.itg.pm.cost.util.ProjectCostUtil:2011/07/03-09:33:05.107 CDT: No TM effort contour found for TM work unit: 1238002 ERROR :ppmLightServiceListenerContainer-1:com.kintana.services:2011/07/03-09:42:37.502 CDT: RM Notification Service did not clear the context from WorkplanTracker and may be loading workplans outside of a transaction ERROR :ppmLightServiceListenerContainer-1:com.kintana.services:2011/07/03-10:40:31.946 CDT: Staffing Profile Period Sum Update Service did not clear the context from WorkplanTracker and may be loading workplans outside of a transaction ERROR :ppmLightServiceListenerContainer-1:com.kintana.services:2011/07/03-11:42:37.490 CDT: RM Notification Service did not clear the context from WorkplanTracker and may be loading workplans outside of a transaction
Re: Miscellaneous errors and comments on SP3 (re-released version)
for the 8.03-bundle, are you using build 1198 or 1208? Check the PPM-information page in the administration-menu for that info.
Regarding the logs:
“No TM effort contour found for TM work unit”
The cost rollup function will distribute the cost of a task over the period of the task length.
When doing so, it tries to get the effort logged in timesheets so that it can distribute the cost as per the effort logged. But if it does not find any time logged for the task which has actual effort against it, it will log an error and distribute the cost evenly over the length of the task.
For example: If a task runs for 3 months and 80 percent of work is done is first month, the only way the service knows this is through timesheets and if it does not find timesheets, it will distribute the cost evenly over the 3 months. This is only done when TM is enabled for the project.
So these errors should have been just a warning rather than an error as the service does the right thing based on available data. This will not affect your data in any way so we can safely ignore this type of error.
"RM Notification Service did not clear the context from WorkplanTracker and may be loading workplans outside of a transaction”
There’s an incident open with RnD that handles reports being executed, but being done so late (scheduled for example 1:10 AM, running 1:42 AM). I can see this string being listed in that incident but am not sure if it’s related. The incident is still open currently. Check your report executions and if you see strange behavior, collect information and log a new ticket.
“Staffing Profile Period Sum Update Service did not clear the context from WorkplanTracker and may be loading workplans outside of a transaction”
Assuming that the above is related to a (code) defect, then the below is related. You can see that both are reporting workplans outside of a transaction. However, both of these error-messages are ‘just’ related to a Service having problems. There SHOULD also be an error reported somewhere related to workplans.
Check the table TM_PENDING_APPROVERS_EVALS for null values – this is one issue I’ve had in the past for the SP Sum Update Service failing with above message, BUT at that time I also had errors with the Task Actual Rollup Service. I believe that there must be more errors in the log. Specifically, I would THEN look for the following string:
java.lang.NumberFormatException: For input string: "null"