Bugs item #2817414, was opened at 2009-07-06 14:29
Message generated for change (Comment added) made by johnvanschie
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=2817414&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: Core
Group: MonetDB4 "stable"
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: John van Schie (johnvanschie)
Assigned to: Peter Boncz (boncz)
Summary: MonetDB allocates >5GB mem while querying a small document

Initial Comment:
OS: Both Fedora Core 10 and Ubuntu 8.04
MonetDB: 4.28.4, compiled from superball (--enable-xquery --enable-optimise), 
64 bit OIDs

Problem:
I have one small document (bookmarks.xml, attached) in MonetDB. 

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

>Comment By: John van Schie (johnvanschie)
Date: 2009-08-12 13:33

Message:
If it is expected behaviour, then this bug can be closed. Still as a user,
I find it hard to understand why querying a 100k document should result in
more than 5GB allocated memory, and a failure to allocate even more memory.

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

Comment By: Sjoerd Mullender (sjoerd)
Date: 2009-08-06 12:55

Message:
While running those 3 queries in parallel loops, the memory consumption of
Mserver grows slowly but seemingly steadily.  When I stop those loops
(actually, I run each query 1000 times) then memory consumption of Mserver
drops again, but not to the same level as it was before the loops started.
However, at the moment I think this (the not coming down completely) can
be explained by memory fragmentation.
Why the memory consumption keeps growing during execution, I don't know. 
Probably because "stuff" cannot be cleaned up after a query because there
are other queries still using the document/collection.

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

Comment By: Martin Kersten (mlkersten)
Date: 2009-07-20 20:42

Message:
Memory allocation by itself is not a problem

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

Comment By: John van Schie (johnvanschie)
Date: 2009-07-06 14:34

Message:
<hit the wrong button, posted too early>
OS: Both Fedora Core 10 and Ubuntu 8.04
MonetDB: 4.28.4, compiled from superball (--enable-xquery
--enable-optimise), 64 bit OIDs

Problem:
I have one small document (bookmarks.xml, attached, 100KB) in MonetDB. 
I query it with 3 mclients simultaneously, with the following commands:

while [ 0 -lt 1 ]; do mclient -lxq -p49255 < /tmp/bookmark-query1.xq
;done
while [ 0 -lt 1 ]; do mclient -lxq -p49255 < /tmp/bookmark-query2.xq
;done
while [ 0 -lt 1 ]; do mclient -lxq -p49255 < /tmp/bookmark-query3.xq
;done

This results in a Mserver process taking more and more amounts of memory,
as if it is leaking memory. I observed it taking up to 5.8 GB of memory,
and then failing with the following error message:

MonetDB>xquery_server: client 6 could not flush prompt
!ERROR: MAPIlisten:mapi_client failed
!OS: Cannot allocate memory
!ERROR: MAPIlisten:mapi_client failed
!OS: Cannot allocate memory
!ERROR: MAPIlisten:mapi_client failed
!OS: Cannot allocate memory

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

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=482468&aid=2817414&group_id=56967

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day 
trial. Simplify your report design, integration and deployment - and focus on 
what you do best, core application coding. Discover what's new with 
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Monetdb-bugs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/monetdb-bugs

Reply via email to