If storage shortage is part of the issue – won't a long session time add to this bottleneck? And following up, what triggers when a session file is deleted? This is all assuming that the piling up of large session files contributes to the problem, which was my understanding before your last comment.

Previously we try to offer allow updating of HP sessuion lifetime, but since it's a lot disable by hoster : we removge the option (a lot of bug report about this).

BUT : we leave it for DB session
manual.limesurvey.org/Optional_settings#...table-based_sessions

Together with our hosting provider, we have now decided to implement the following measures:

• Optimizing the survey design
• Adjusting the PHP gargabe collector
• Changing to table-based sessions
• Adjusting the session lifetime for table-based sessions

Hopefully, this will prevent the problem in the future. Thanks again for your help.

Update: We have since changed our sessions to table based sessions. On a running survey with a large number of participants, the server collapsed almost instantly (within two minutes). Our hosting provider identified the problem: session entries are deleted from lime_sessions based on the value in the column 'expire'. Since this column did not have an index, the request took longer and longer as the session table was growing, blocking the table for other requests. By creating an index on the column 'expire', the issue could be resolved.

I realize that this is only an issue when there is a huge amounts of entries in the session table. However: could it be an option to make it the standard way for LS to create the session table with an index on 'expire'?

Yes , sure , can you report the issue please :
community.limesurvey.org/bug-tracker/