[Antelope] Question dbdetect

[Michael West]

Hi Lepolt,

I assume someone else will chime on whether there is a hard file size
limit at 1022 MB.

My comment is that regardless of a specific limit you may find it
easier to break your time ranges into pieces. A 1 Gb detection table
would be in excess of 5M records. As you move further downstream in
your processing you will find that managing resources (both space and
CPU) will be easier if you split your data into something like monthly
catalogs or smaller. There are several unanticipated benefits such as
allowing you to farm CPU-intensive jobs to multiple machines in
parallel, should you choose. You can always cat specific tables back
into larger time ranges. There are several tools specifically for
managing this type of data such as dbconcat_events and dbcentral.

While there are downsides, I think you will find that for one-off
projects, the philosophy of several smaller databases will serve you
well. Use the -tstart and -twin flags for dbdetect. If you want
specific examples feel free to contact me directly.

- Mike West


-- 
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     Michael West
 Geophysical Institute
Alaska Volcano Observatory
Univ. of Alaska Fairbanks
Fairbanks, AK 99775
907-474-6977 (office)
907-687-4288 (cell)
www.giseis.alaska.edu/input/west
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~






> Hello,
>
> My name is Lepolt Linkimer, I am a PhD student at the Department of
> Geosciences at the University of Arizona. I am in the process of detecting
> and associating local earthquakes using data from both the SIEMBRA and ESP
> experiments in the Argentinean Andes. I have a total of 52 stations and the
> seismicity in this region is *very high*. I am interested in detecting
> possibly all local earthquakes. I have been having the following error when
> running dbdetect for the entire dataset:
>
> command line: dbdetect -v -pf /arc6/lepoltl/dbdetect.pf SIEMBRA_db
> SIEMBRA_db
>
> Attached the pf file I am using.
>
> Error:
> ------------
> dbdetect: Can't mmap table 'SIEMBRA_ESP_db.detection'.
>         Cannot allocate memory
> dbdetect: dboutdetection: dbaddv() error.
> Detection OFF at 12/01/2008  16:23:09.675 for        LUNA/BHE:      BW 1.0 4
> 5.0 4     18.525 at 12/01/2008  16:23:06.800
>
> dbdetect: Segmentation fault
> dbdetect *fault*: Signal SIGSEGV : Segmentation fault
> Abort
> -----------
>
> I am working on a Mac. My operating system:
> $ uname -a
> Darwin arc.geo.Arizona.EDU 9.8.0 Darwin Kernel Version 9.8.0: Wed Jul 15
> 16:55:01 PDT 2009; root:xnu-1228.15.4~1/RELEASE_I386 i386
>
> I am using antelope 4.10 and I believe all patches have been applied.
>
> The error occurs when the output file "SIEMBRA_ESP_db.detection" reaches a
> size of exactly 1022Mb. Is there any limit of filesize for the output? Is
> this file size "normal". I have decided to run the detections by periods of
> time and I ended up with 8 detection tables (the project is two years).
> dbdetect works fine for me unless the output reaches 1022 Mb. I am now
> running the associations (bdgrassoc) for each detection table but I am
> wondering if this approach will give me complications in the future. What do
> you think?
>
> Thanks for your help!
>
> Lepolt
> --
> PhD Candidate
> Department of Geosciences
> University of Arizona
> http://www.geo.arizona.edu/~linkimer/
>
> _______________________________________________
> Antelope mailing list
> Antelope at brtt.net
> http://brtt.net/mailman/listinfo/antelope_brtt.net
>
>