Application crash when running the binary for a few days on some systems
Description
When running DATAGERRY from the binary, on some systems the application stopped working after a few days.
Reason:
When starting the binary, all required files were extracted to /tmp directory. systemd-tmpfiles may delete files from that directory, that are older than x days. This will make the DATAGERRY WebUI unresponsive.
Environment
Activity
The following options exists to solve the problem:
create a configuration file, which excludes the temporary directory created by PyInstaller from cleanup
/usr/lib/tmpfiles.d/datagerry.conf
Use another directory to let PyInstaller extract the content of the binary:
TMPDIR=/root ./datagerry -c ./cmdb.conf -s
The following configuration file for systemd-tmpfile in /usr/lib/tmpfiles.d/datagerry.conf will exclude PyInstaller directories from cleanup:
Configuration added to RPM and setup script. Needs to be tested with in different environments:
Rollout of RPM on CentOS/RHEL
Rollout of TAR.GZ on Debian/Ubuntu
Check, if Docker Image still works correctly
Tested: Rollout RPM on CentOS 7 & 8
Tested: Rollout TAR on Ubuntu 20.04.1 LTS & Ubuntu 20.10
I also tried Docker. Do not see any problems. Ran through.
tmpfiles were created, tmpfiles.d service did not delete any files when moving the date over the configured time.
Assignee
Reporter
Labels
Tester
Sprint
Fix versions
Priority
