Accept Changsets very slow, huge derby database?
We've been running Jazz for perhaps 8 months now, and the database, still based in Derby, has been getting huge: at this point it's about 4GB.
At the same time, accepting simple changesets now takes over 4 minutes. The java teamserver process consumes 197% CPU (dual-core system) and takes about 1GB RAM. We see very little disk I/O. Our dev looked at the database and saw it contains nearly 1.5 billion rows in one table!
How can we fix this problem, and is there some kind of ongoing maintenance we've been neglecting to do?
Thank you very much!
At the same time, accepting simple changesets now takes over 4 minutes. The java teamserver process consumes 197% CPU (dual-core system) and takes about 1GB RAM. We see very little disk I/O. Our dev looked at the database and saw it contains nearly 1.5 billion rows in one table!
How can we fix this problem, and is there some kind of ongoing maintenance we've been neglecting to do?
Thank you very much!
One answer
It looks like you reached the limits of Derby. Which table has 1.5 billion rows? I'd love to see details about how many artifacts your have in this repository, could you send us the output of the repository sizing reports, for example:
https://<yourserver>:9443/jazz/web/projects/Jazz%20Project#action=com.ibm.team.reports.viewQuery&queryUUID=_cEF7UHREEdywet_K-pGQGw&name=Latest%20Metrics
You can browse the reports and see how many files/folders/work items you actually have in the repository.
If you read the deployment guide, we really encourage teams with databases of this size to use an industrial DB. Derby was provided as a getting started solution, but not as a long term solution for data of this size.
https://jazz.net/learn/LearnItem.jsp?href=content/tech-notes/jazz-team-server-0_6-deployment-guide/index.html
You can download the express edition and migrate to DB2. In our own selfhosting the size of our own repository is described in:
https://jazz.net/wiki/bin/view/Main/PerformanceScalability#Selfhosting_Repository_data_sets
You can use this to compare with your dataset.
Jean-Michel
https://<yourserver>:9443/jazz/web/projects/Jazz%20Project#action=com.ibm.team.reports.viewQuery&queryUUID=_cEF7UHREEdywet_K-pGQGw&name=Latest%20Metrics
You can browse the reports and see how many files/folders/work items you actually have in the repository.
If you read the deployment guide, we really encourage teams with databases of this size to use an industrial DB. Derby was provided as a getting started solution, but not as a long term solution for data of this size.
https://jazz.net/learn/LearnItem.jsp?href=content/tech-notes/jazz-team-server-0_6-deployment-guide/index.html
You can download the express edition and migrate to DB2. In our own selfhosting the size of our own repository is described in:
https://jazz.net/wiki/bin/view/Main/PerformanceScalability#Selfhosting_Repository_data_sets
You can use this to compare with your dataset.
Jean-Michel