Open matthias-ronge opened 3 years ago
@matthias-ronge: Could this be the reason for the following issue?
The behaviour of #4760 occurs in our test system with more then 400.000 processes. In the preview system the creation of newspaper processes is not slow.
However, the general check for duplicate processes must be possible before newspaper processes are created!
@matthias-ronge: Could this be the reason for the following issue?
• Performance problem: Creating newspaper processes - creation of processes is very slow #4760
The behaviour of #4760 occurs in our test system with more then 400.000 processes. In the preview system the creation of newspaper processes is not slow.
No idea; but when process creation slows down with the total number of processes, we should pay attention to the database queries and index actions that are taking place. If, for each newly created process, the project is re-indexed, and the meta.xml
is parsed for each process of the project—or something comparable—then these actions add up.
@matthias-ronge : Is this still a problem or has it been solved by the following pull request?
There is a lot of checking for the same files exist again and again when creating newspapers. File system access is comparably slow in Java and should be avoided to happen again and again.
Goal: For known fils, don’t check existence.