Description
If you notice over time the CloudBees CD (CloudBees Flow) Web server response is becoming slower or you discover that the response is unacceptable, something has changed.
-
What could cause the poor response? Is the remedy external to CloudBees CD (CloudBees Flow)?
-
What CloudBees CD (CloudBees Flow) files can I look at to help diagnose determine the source of the poor response?
Specific areas of slow response can include:
-
login takes more that a couple of seconds
-
browsing changing between CloudBees CD (CloudBees Flow) tabs, note Home and Search tab response
-
browsing projects/procedures
-
browsing jobs
-
browsing job details
-
browsing a job’s log files
-
using the GUI to delete jobs
-
a step is marked as waiting for resource when resource is available
-
a step’s execution time
-
starting a job
Solutions
Changing between CloudBees CD (CloudBees Flow) tabs - name which are slow
Symptom
When selecting tabs on the CloudBees CD (CloudBees Flow) GUI, the response is sluggish. Select the tabs in the following order:
-
search tab - there is no database query to bring up the tab and the response is measuring the Web Server Response only, no CloudBees CD (CloudBees Flow) Server use.
-
the project tab - simple query to the CloudBees CD (CloudBees Flow) Server.
-
the jobs tab - could be the largest single table on large installations.
-
the home page - can be the most complex page to build and reference a possibly unlimited number of user properties references.
Possible Causes
-
Database connection port is blocked.
-
DNS resolutions is lengthy
Remedies
-
Ping the DNS name.
-
Talk to IT about firewall port restrictions.
Job details display
Symptom
It takes over half a minute to display a details of a job.
Possible Causes
-
The workspace directory file system has thousands of file.
-
Network access to the workspace file system from the Web server is slow - low bandwidth.
Remedies
-
Delete jobs with workspaces on the slow performing file system.
-
Install a Web server in the workspace file system’s domain to address low bandwidth.
Login
Symptom
Login time is more than a few seconds.
Possible Causes
-
A directory provider search takes a long time to contact all nodes in a corporate setup.
-
One or more of the providers cannot be contacted because of network configurations. The failure to contact timeout can be lengthy.
Remedies
-
Use JXplorer to examine directory provider configurations.
step is marked as waiting for resource
Symptom
The job details display has a white icon meaning the step is ready to run but not yet started.
Possible Causes
-
The CloudBees CD (CloudBees Flow) database step table is very large.
-
the database is not tuned for step table queries
-
Workspace may not be available
Remedies
-
Prune jobs to reduce the size of the step table.
-
use database tuning tools like Oracle Enterprise Manager improve the query performance.
-
Drill to the step details page to see if the cause is workspace availability
Procedure/project display
Symptom
From the browser, there is a delay when selecting a project/procedure.
Possible Cause
The database is busy serving other requests:
-
exports/imports are in process
-
a large number of job deletes are occurring, such as when deleting a project, which deletes all projects jobs, or when pruning jobs
Remedies
-
Use the commander.log files to determine the obstructing database activity.
LazyInitialization Exception
Symptom
Logs contain these exceptions or the UI displays exceptions when trying to access job related information.
Possible Cause
The database is busy serving other requests:
-
exports/imports are in process
-
a large number of job deletes are occurring, such as when deleting a project, which deletes all projects jobs, or when pruning jobs
-
Memory is at its peak usage
Remedies
-
Use the commander.log files to determine the obstructing database activity.
-
Send log files to EC support to analyze memory configuration and information
-
Check conf/wrapper.conf file to make sure sufficient memory has been allowed