carlospzurita
commented
5 years ago One exception that we keep encountering is this one
ERROR d.i.e.w.c.TestResultController - Report not found:
de.interactive_instruments.exceptions.ObjectWithIdNotFoundException: TestRunDao: An object with the id "8c8d8fcf-a732-4adf-b95f-da65406f1f63" could not be found!Also, we noticed that the ffiles in /etf/ds/attachments and /etf/testdata were not emptied, so we set up a crontab on the cloud instance to execute the cleanup operation.
sudo find /efs/ds/obj -type f -name 'TestObject-*' -mtime +7 -exec rm {} \;
sudo find /efs/ds/obj -type f -name 'TestRun-*' -mtime +7 -exec rm {} \;
sudo find /efs/ds/obj -type f -name 'TestTaskResult-*' -mtime +7 -exec rm {} \;
sudo find /efs/ds/attachments -name '*' -mtime +7 -exec rm '-rf' {} \;
sudo find /efs/testdata -name '*' -mtime +7 -exec rm '-rf' {} \;¿Maybe the TestRuns references were kept on the database, or the Java cache, but deleted from the datastore, causing the exception and preventing from executing the cleanup of the rest of the files?
jonherrmann
Description
On the production deployment of the INSPIRE validator, there has been detected a problem with the periodic cleanup operations of test data. This operations take place every day at midnight, and checks if there are test reports and associated files older than the parameters
etf.testobject.temporary.lifetime.expirationandetf.testreports.lifetime.expirationon the ETF properties file.This operation is not executed completely, some files are left behind. We have identified the following files on the ETF directories:
ds/attachmentsTestObject-*.xmlonds/objtestdataAll these directories and files were created using the user
jetty, and the owner and the group have permission to read and write on them.The size of the ETF file system increases considerably because of this problem, and any operation (listing test-reports, restart of the instance) takes more and more time to execute.
Operating systems and browser
The issue is happening on an AWS EC2 instance with Amazon Linux/2.12.14. The validator runs on a Docker container inside this instance. The application server being used is Jetty 9.3.6. The
/etfdirectory is created as a volume inside the container, mapped to an external AWS storage services mounted as a NFS device.This issue was also being reported on the old sanbox instance from the JRC, deployed on a Linux-based OS, and using Tomcat as an application server.
Steps to Reproduce
[Describe the exact steps which can be used to reproduce the problem]
diff -qrbetween a copy of the ETF file system before cleanup, and after.Expected behavior: [Which behavior you expected to see instead and why.]
The described files should be removed from the file system, given that their TestRun have been removed.
Actual behavior: [Describe the behavior you observed after following the steps]
The files are kept on the file system.