directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Emmanuel Lecharny (JIRA)" <>
Subject [jira] Created: (DIRSERVER-1527) Subtree processing has to be completely rethought
Date Thu, 15 Jul 2010 15:58:53 GMT
Subtree processing has to be completely rethought

                 Key: DIRSERVER-1527
             Project: Directory ApacheDS
          Issue Type: Bug
            Reporter: Emmanuel Lecharny
            Priority: Critical
             Fix For: 2.0.0-RC1

Currently, managing subtrees is done by adding the subentry, and then modify all the entries
selected by the subtreeSpecifciation by adding a new attribute pointing to the assciated subentry.
This is simply overkilling : with millions of entries, that means we may have to modify millions
of them, which will cost a hell of a time, assuming that the server is capable of only 2000
modifications per second (that's almost 10 minutes for 1M entries to be modified...).

We *have* to manage subtrees differently. One option would be to filter the entries by evaluating
them when they are manipulated, everywhere we use subtrees :
- Schema (not used yet)
- Triggers
- CollectiveAttributes

Of course, it will slow down a bit those operations, but we won't see the server stopped for
minutes while processing a subtree addition, removal or modification.

Plus the fact that some of the current operations are not handled correctly, leaving invalid
subentry references into many entries... (MoveAndRename for instance : the old ref remains
in entries)

This message is automatically generated by JIRA.
You can reply to this email to add a comment to the issue online.

View raw message