[prev in list] [next in list] [prev in thread] [next in thread] 

List:       novalug
Subject:    Re: [novalug] updatedb question
From:       "C. Megan Larko" <megan () rainfall4 ! gsfc ! nasa ! gov>
Date:       2005-08-29 2:08:12
Message-ID: 20050829020812.GC30603 () rainfall4 ! gsfc ! nasa ! gov
[Download RAW message or body]

Megan here.  My $.02 comments in-line.

On Sun, Aug 28, 2005 at 11:37:01AM -0400, Clif Flynt wrote:
> On Sat, Aug 27, 2005 at 10:58:38PM -0400, Bonnie Dalzell wrote:
> > 
> > PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
> > 8836 root      39  19 56792  54m  480 R 94.7 10.8 567:33.11 updatedb
> > ...
> > is PR the task priority?
> 
>   Yes
> > how do you set a task priority?
> 
>   Check the manpage for "nice".  I don't recall if you can modify the 
> priority of an existing program.  You may want to add "nice" to the
> startups cript.
> 
> > does a task owned by root have higher priority than one started by a user?
> 
>   No.

One may renice a root program by becomming root and using the renice
command.  Negative values are a higher priority.  One doesn't want to
go too negative (-16 or closer to -20) because then the program may
fight with lower-level, more critical system tasks.  I generally may
take a program to a -10.   Personally I have not had a problem with
the up2date job.  

One may (and I have) requested up2date to "-e" or exclude certain
paths.  A 16Tb Attabeast would be a good mount point to exclude...

megan
> 
> > how long can I expect ubdatedb to tak? I am planning to let it run
> > overnight.
> 
>   I've got about 300 Gig of disks on a 1Ghz PIII box.  updatedb runs in
> about an hour.  I suspect you've got something slowing down your I/O.
> Are some of your disks NFS mounted, or sitting on USB?  Any chance of a
> failing drive that's taking multiple seeks and resets to read some
> directory tracks?
> 
>   The memory usage looks large to me.  I don't recall if updatedb
> recognizes (and avoids) symlink loops like:
> 
>   ln -s /foo /foo/bar/baz 
> 
>   If you do a 'locate *' you'll get a list of every file updatedb has
> found on your system.  You might see huge paths that would give you
> a clue about a loop.
> 
> > why does it take so many of the  cpu resources?
> 
>   Updatedb is building a database with some hash tables and such.  That
> shouldn't be a lot of CPU activity.  Here's what updatedb looks like
> on my system:
> 
>   PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND           
>  2564 root      18   0  3000  652  428 D  5.3  0.1   0:01.07 updatedb          
> 
> Clif
> 
> --
> .... Clif Flynt ... http://www.cflynt.com ... clif@cflynt.com ...
> ..Tcl/Tk: A Developer's Guide (2'nd edition) - Morgan Kauffman ..
> ..12th Annual Tcl/Tk Conference:  Oct 24-28, 2005,  Portland, OR.
> .............  http://www.tcl.tk/community/tcl2005/  ............
> 
> 
> 
> _______________________________________________
> novalug mailing list
> novalug@tux.org
> http://www.tux.org/mailman/listinfo/novalug
> for subscribe/unsubscribe see web page

-- 
-----------------------------------------------------------------
"God, grant me the serenity to prioritize the things I cannot delegate,
courage to say no when I need to, and wisdom to know when to go home."
                            -Plagiarized from an unknown source
-----------------------------------------------------------------
C. Megan Larko
Laboratory for Hydrospheric Sciences
Code 614.3 Goddard Space Flight Center
Greenbelt, Maryland 20771
_______________________________________________
novalug mailing list
novalug@tux.org
http://www.tux.org/mailman/listinfo/novalug
for subscribe/unsubscribe see web page
[prev in list] [next in list] [prev in thread] [next in thread]