Hi, Deepa,

Sorry, looks like this message somehow got neglected..

Here is a suggestion for tracking down the memory usage.

-- add '-v 10' to your command line for ibis

you should be able to see Warning message before the crash about
requesting a large amounts of memory.

Roughly, if you have two tables with 1 million rows each, and you want
to generate a full cross product (e.g., a join), you have to be
prepared to deal with 1 million times 1 million rows.  Looks like you
are providing some restrictions on the tables, which should reduce the
number of rows in the cross-product table.  If you can run the
selection conditions on each table separately, you should be able to
get a upper bound on the number of rows in the cross product table.

Hope this helps.

john

On 5/25/12 1:49 PM, Deepa Sahni wrote:
> The number of rows in one million. I used ardea utility to load the
> data from a .csv file.
> 
> Examples of queries I ran are given below(I am including the output
> because it includes the query too ). Although from what I could see
> any query with these columns in the where clause is failing.
> 
> ./benchmark_q2.sh: line 46:  1923 Killed                 
> /home/deepa/Downloads/fastbit-ibis1.3.0/examples/ibis -d
> /home/deepa/Downloads/fastbit-ibis1.3.0/tmp -q "select count(*) where 
> K2=2 and K4=3"
> 
> 
>  ./benchmark_q2.sh: line 48:  1963 Killed                 
> /home/deepa/Downloads/fastbit-ibis1.3.0/examples/ibis -d
> /home/deepa/Downloads/fastbit-ibis1.3.0/tmp -q "select count(*) where 
> K2=2 and K10=3"
> 
> 
> ./benchmark_q2.sh: line 50:  2019 Segmentation fault     
> /home/deepa/Downloads/fastbit-ibis1.3.0/examples/ibis -d
> /home/deepa/Downloads/fastbit-ibis1.3.0/tmp -q "select count(*) where 
> K2=2 and K100=3"
> 
> Also if you think it does need more memory, could you give some
> specifics. In the sense if I want to fix the issues how much more
> memory should I make available to it. Does it need more main memory or
> hard disk space?
> 
> Thanks,
> Deepa
> ./benchmark_q2.sh: line 52:  2047 Segmentation fault     
> /home/deepa/Downloads/fastbit-ibis1.3.0/examples/ibis -d
> /home/deepa/Downloads/fastbit-ibis1.3.0/tmp -q "select count(*) where 
> K2=2 and K10K=3"
> 
> 
> On Fri, May 25, 2012 at 1:20 PM, K. John Wu <[email protected]
> <mailto:[email protected]>> wrote:
> 
>     Hi, Deepa,
> 
>     I would guess that you are using data generated by setqgen.  Would you
>     mind tell us how many rows are there?  What are the exact queries
>     you've used?
> 
>     I would assume that your join is a little too big to be held in 1GB of
>     memory.  If you give me the about information, I would be able to
>     confirm this speculation.
> 
>     John
> 
> 
> 
>     On 5/25/12 1:07 PM, Deepa Sahni wrote:
>     >
>     > Hi John,
>     >
>     > I have attached the data partition details of all the columns. The
>     > issue I am having is with the following columns.
>     >
>     > Column Name                  Issue
>     > --------------------------------------------------
>     > k4                                   killed
>     > k10                                 killed
>     > k10k                               segmentation fault
>     > k100                               segmentation fault
>     >
>     > In addition to this I am running FastBit on ubuntu(running on top of
>     > VMPlayer) and the ram available to it is 1GB and hard disk is 20GB.
>     > If you need any other information I will provide that too.
>     >
>     > Thanks,
>     > Deepa
>     >
>     >
>     >
>     > On Fri, May 25, 2012 at 9:36 AM, K. John Wu <[email protected]
>     <mailto:[email protected]>
>     > <mailto:[email protected] <mailto:[email protected]>>> wrote:
>     >
>     >     Yes, you are probably encounter memory shortage.  Since you
>     did not
>     >     give any size information, there is no way for me to verify your
>     >     statement;-)
>     >
>     >     John
>     >
>     >
>     >     On 5/25/12 1:09 AM, Deepa Sahni wrote:
>     >     > Hi,
>     >     >
>     >     >  I need some help debugging an issue. FYI I am running
>     FastBit on
>     >     > ubuntu.
>     >     >
>     >     > Whenever I run queries with certain columns in the where
>     clause
>     >     I get
>     >     > the following errors.
>     >     >
>     >     > for column K100 : Process killed
>     >     > for column K10 :Segmentation Fault
>     >     >
>     >     > I did not create indexes on any of the columns, since my
>     >     understanding
>     >     > was that FastBit creates indexes as and when needed. As I
>     ran my
>     >     > queries indexes got created. But for the columns mentioned
>     above I
>     >     > repeatedly get the same error for all select statements
>     with these
>     >     > columns in the where clause.
>     >     >
>     >     > I tried deleting all the other indexes to clear up some
>     memory for
>     >     > these(K100, K10) column indexes(I think this might be
>     happening
>     >     > because of memory shortage). It did not fix the problem.
>     Could some
>     >     > shed some light on what might be causing this error or how to
>     >     debug or
>     >     > fix this.
>     >     >
>     >     >
>     >     >         Thanks in advance,
>     >     >         Deepa
>     >     >
>     >     >
>     >     >
>     >     >
>     >     >
>     >     > _______________________________________________
>     >     > FastBit-users mailing list
>     >     > [email protected]
>     <mailto:[email protected]>
>     <mailto:[email protected]
>     <mailto:[email protected]>>
>     >     > https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users
>     >
>     >
> 
> 
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to