Bugs item #1650205, was opened at 2007-02-02 00:03
Message generated for change (Comment added) made by boncz
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=1650205&group_id=56967

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: PF/runtime
Group: (zombie: Pathfinder 0.16)
>Status: Pending
Resolution: Fixed
Priority: 3
Private: No
Submitted By: Peter Boncz (boncz)
Assigned to: Sjoerd Mullender (sjoerd)
Summary: PF: empty pages never reused

Initial Comment:
I was toying a bit with bulk updates

/data/home/boncz/bench/mod.xq

doing mod:addbulk() followed by mod:del()

thus, insert 14 pages, and then delete all those nodes again. if done 
repeatedly, one notices that the document keeps growing by 14 pages each update.

that is, new pages are allocated; the data on them is subsequently deleted, but 
the pages are not freed, *nor* are they re-used.

it turns out that the insert code just checks whether the hole *on-the-page* is 
sufficiently large. As the insert requires 14 pages, this is never the case. 
Therefore, it decides to insert new pages.

Continuing along these lines, if our insert point happens to be very near the 
page boundary (say fully at the end), and we have an access pattern that 
reatedly inserts "as first" there, each such insert of a single node will cause 
the insertion of a new page!!

thus, I would propose the following:
- in deletes, reclaim fully free pages
- if less than a full new page is missing, check if the next still not 
reclaimed old page has enough free space (at the end). Use movedata to make 
room on that page.

that should strongly improve the fill degree of the rid table in many scenarios.


----------------------------------------------------------------------

>Comment By: Peter Boncz (boncz)
Date: 2009-04-07 10:31

Message:
will not be fixed, will disappear eventually in the new MonetDB5 runtime

----------------------------------------------------------------------

Comment By: Stefan Manegold (stmane)
Date: 2008-02-10 14:18

Message:
Logged In: YES 
user_id=572415
Originator: NO

tagged as "PF" in subject

----------------------------------------------------------------------

Comment By: Stefan Manegold (stmane)
Date: 2007-06-04 20:40

Message:
Logged In: YES 
user_id=572415
Originator: NO

Peter, Sjoerd,

is this one still an open issues after all the recent XQUF code
changes/improvements?

Stefan


----------------------------------------------------------------------

Comment By: Sjoerd Mullender (sjoerd)
Date: 2007-02-19 13:12

Message:
Logged In: YES 
user_id=43607
Originator: NO

Lowered priority.

----------------------------------------------------------------------

Comment By: Sjoerd Mullender (sjoerd)
Date: 2007-02-13 09:56

Message:
Logged In: YES 
user_id=43607
Originator: NO

Newly empty pages are now returned to the system.

The other part of the proposal needs more thought and work, so I'll keep
this report open.

----------------------------------------------------------------------

Comment By: Martin Kersten (mlkersten)
Date: 2007-02-09 18:14

Message:
Logged In: YES 
user_id=490798
Originator: NO

tag for filtering

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=1650205&group_id=56967

------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com
_______________________________________________
Monetdb-bugs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/monetdb-bugs

Reply via email to