XML backups get stuck

[eloiferrer]

Describe the bug
The XML backup script just gets stuck or takes forever to finish with no apparent output. Filtering for namespaces does not help.

For instance:

/usr/local/bin/php /var/www/html/maintenance/dumpBackup.php --current --output=gzip:/data/xml_backup_test.gz --filter=namespace:0 --conf /shared/LocalSettings.php

just hangs there, showing now output. Only after waiting several minutes it shows the process output for 1000 pages and an estimated finishing time > 1 month. Then it hangs again.

Expected behavior
(A clear and concise description of what you expected to happen.)

To Reproduce
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

Screenshots
(If applicable, add screenshots to help explain your problem.)

Additional context
Add any other context about the problem here.

• In example, information about the device used for producing the bug

Checklist for this issue:
(Some checks for making sure this issue is completely formulated)

• Assignee has been set for this issue
• All fields of the issue have been filled
• the main MaRDI project has been assigned as project

[eloiferrer]

If we don't need the XML backups I suggest we deactivate them permanently.
Regardless of that, we can create JSON backups of only the Wikibase entities.

[physikerwelt]

I would first try to exclude items and properties as well and test again.

[eloiferrer]

That is what I have done. Even trying with much more restrictive filters such as --filter=namespace:0 which should only include the main namespace, gets stuck.

[physikerwelt]

@eloiferrer did you see the following

Only after waiting several minutes it shows the process output for 1000 pages

for ns:0?

[eloiferrer]

Yes. For ns:0 and also with any other configuration I've tried. The script seems to get stuck from the very beginning, no matter the options. It seems that the script goes through all the pages anyway, and gives some output as these get processed.

[physikerwelt]

I would like to understand why this is the case. There should be less than 1000 pages in ns:0. So it should finish in one hour or less.

[physikerwelt]

I found my own docu on https://www.mediawiki.org/wiki/Manual:DumpBackup.php :-) and I think filter should go to the query but plugin would still process every page.

[physikerwelt]

@eloiferrer you are right. The namespace filter is not in the query. https://gerrit.wikimedia.org/r/c/mediawiki/core/+/504792 while I share the view of @brightbyte. Espcially with our memory limit for the single database setup, it will be impossible to run this daily. One option to get around would be to rewrite the query via the hook.
I think given the effort, we should exclude XML backups from the daily runs.