Question regarding data directory structure on upcoming Elgg 1.9 / bucket_size

I just began testing Elgg 1.9 regarding plugin compatibility etc.

One thing I noticed (saw it mentioned before on github) is the new directory structure within the data directory. Am I right in assuming that there a separate "top-level" directories (1, 2, 3, etc.) for every 5000 users? Does this mean that there could be in theory up to 5000 user directories (guid-named) within each of the top-level directories.

This could be quite difficult / impossible to handle with any ftp client (and maybe also some server based file browsers). At least the ftp clients I know of can't display more than 1998 files within a single directory. If they even fail to copy / move any files above this limit you can't remote-manage your data directory anymore at all!

Am I wrong in my concern that the bucket_size of 5000 could result in problems? May I ask what's been the reason for restructuring the data directory again?

  • The reason for the restructuring I believe is to remove the element of time from the structure.  Previously the location of files was dependent on entity.time_created - but this caused problems for people who moved servers with different timezones - the timezone could offset the effective date of the time_created such that the files were no longer in the right places.  Having it by guid with no dependence on time interpretation is much better, though I don't know enough about the bucket_size issue you bring up to answer that.

  • Interesting. Any other change that we must be aware of?

     

  • Oh oh... Why is there any RSS in this 1.9 ?