We just upgraded our production environment to PPM 9.14 from 9.11 last week. Since then we are seeing strange and random behavior with the new table component object. The issue is that some user’s updates to table component data are not being saved. There are no error messages presented to the user or any error messages in the serverlog.txt files. Everything looks normal except the data is missing. Simple fields are saving fine all the time. Here is what we are seeing:
1: Some users have no issues. Saves work fine.
2: Some users can save in one browser vs. another. More times than not, Firefox 16 works, and IE 8 does not.
3: Sometime it just starts working for a user after a few days.
4: Some users can save one record but not another using the same browser.
As I said, very strange and random. As far as we can tell nothing has changed on their PCs. We have a hard test case where the user can update data fine on the QA system but not on the Production system. Both as running 9.14 and were upgraded the same way. So we don’t think it client issue. We had no issues prior to 9.14 install on our production system.
We are working HP support but I thought I would ask here for suggestion of things to try or how to debug.
Is this happening with all the requests using table components? Is this happening only with exsiting requests? (Created prior the upgrade) Have you tried to delete the server cache? (work and tmp folder)? Also, make sure you deleted the java cache from all the machines
1. Stop PPM server. 2. Delete work & tmp folders from PPM_HOME/server/SERVER_NAME (repeat for all nodes) 3. From the command prompt under PPM_HOME/bin run: sh kJSPCompiler.sh 4. Re-start PPM server 5. From the command prompt under PPM_HOME/bin run: sh kRunCacheManager.sh 6. Clean browser and java cache from the machines.
Dayana, I am confused about one point in your response. My understanding is that the caches are maintained in each node's memory space, in the Java heap. The current contents are reported in the Cache Manager Statistics report. When a node is stopped, the cache is released along with the rest of the process's memory and the cache will be empty when the node is restarted. What is accomplished by running kRunCacheManager.sh when the application is stopped or has just been started? What am I missing here?
--remember to kudos people who helped solve your problem
Most of the time if a user has a problem it affects all requests that user tries to update. It is happening on new and existing requests. We have already did all the steps you listed below a couple of times now except for the java cache suggestion. I assume you are talking about on the clearing java cache on client (PC) side .
One of our hard test cases for failure started working yesterday. We were going to test the java cache suggestion so I had her clear her IE8 cache and try the request to make sure it stilled failed. Guess what, it started working for her. I am not sure if it was working before she cleared the cache or if clearing the browser cache fixed it. We did have the user clear her cache multiple times over the last few days and that did not correct the issue.
A couple of users have reported that if they clear the browser cache (my manager is one of them J) that it starts working for them. It appears that process doesn’t work all the time. We will have the user clear the java temp files the next problem ticket that comes in for this problem if just clearing browser cache does not work.