In our case, unfortunately, no.

Each partition contains data captured at intervals of five minutes that are
stored in the directory whose name is: <year>/<month>/<day>/<hour>/<minute>,
then the partitions are repeated.

There is any simple solution?

Valeria

2010/5/15 K. John Wu <[email protected]>

> Hi, Valeria,
>
> I see your point about the data actually being different, however, it
> would be a lot easier for us if we can rely on the fact that the data
> partitions are named uniquely.  If the names are not unique, it would
> be harder for us to handle the queries with a FROM clause.
>
> Is there any chance you can assign a different name to each data
> partition?
>
> Thanks.
>
> John
>
>
> On 5/15/2010 8:41 AM, Valeria Lorenzetti wrote:
> > Hi John,
> >
> > I think the library doesn't properly handle multiple directories that
> > have "same name", like in this case:
> >
> > first partition: database/13/00
> > second partition: database/14/00
> >
> > These directories are two different partitions, which contain different
> > data and different rows, as you can see from the files -part.txt:
> >
> >  > more database/13/00/-part.txt
> > BEGIN HEADER
> > Name = "00"
> > Description = "Data partition .."
> > Number_of_columns = 12
> > Number_of_rows = 1425408
> > Timestamp = 1255388390
> > END HEADER
> >
> >  > more database/14/00/-part.txt
> > BEGIN HEADER
> > Name = 00
> > Description = "Data partition .."
> > Number_of_rows = 663552
> > Number_of_columns = 12
> > Timestamp = 1255399170
> > END HEADER
> >
> > If I perform a query like that:
> >
> > thula -s "L4_DST_PORT" -d database/13/00  -d database/14/00 -w
> > "L4_SRC_PORT=101"
> > gatherParts -- replacing the old partition named 00 with new data
> > partition from database/14/00
> >
> > FastBit believes to use the same partition twice, and the query is
> > executed only on the last partition (which is wrong)!
> >
> > Looking in the code (fastbit-ibis1.1.8/src/part.cpp line 12475), I think
> > you are checking the name of partition (that may be not unique), while I
> > think it would be correct to check the timestamp.
> >
> > Let me know what you think,
> > best regards,
> >
> > Valeria
> >
> >
> >
> > _______________________________________________
> > FastBit-users mailing list
> > [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
>
_______________________________________________
FastBit-users mailing list
[email protected]
https://hpcrdm.lbl.gov/cgi-bin/mailman/listinfo/fastbit-users

Reply via email to