Re: [ferret_users] Error signal 11

[Russ Fiedler <russell.fiedler@xxxxxxxx>]

Hi Sarah,

This signal indicates that there is a memory reference problem. If you type

man 7 signal

on you machine you can see the various signals.

In your case I think you are requesting more memory than your system has readily available and Ferret is running into problems. I guess it keeps filling its memory cache until it hits the limit and then start to flush it.

You are requesting 20K megawords of memory which is equal to 80GB! Quite a bit and I guess is way more than you have available. Try asking for more reasonable amount.

Also if this is a large 3/4D calculation it may be more suitable (and considerably quicker) to break up the calculation into smaller amounts.
Say just a few latitudes at a time and save the results to disk or save sigma at each time step and then read it back in order to calculate the time average.

Russ



On 13/11/14 08:48, Sarah Marcil wrote:

Hi,

   I'm experiencing some trouble with Ferret since a week or two and I can't figure out what is the main problem. When I try to run a script, Ferret seems to crash with this error message:  **ERROR Ferret crash; signal = 11

For the same script, sometimes it works and some other time it doesn't.

Does anyone know what signal=11 is? What can cause the crash?

Here is an example of the script, I'm trying to calculate the mean density in the mixed layer depth.

yes? cancel data/all

yes? cancel view

yes? SET MEMORY/SIZE=20000

 Cached data cleared from memory

yes? let sigma0_diff_10_pi = sigma0[d=5] - 0.5*(sigma0[d=5,k=1]+sigma0[d=5,k=2])

yes? let mld_dbm_pi = if( sigma0_diff_10_pi[k=@max] GT 0.03 ) then sigma0_diff_10_pi[z=@loc:0.03] else deptho[d=4]

yes? let mld_1000= if (mld_dbm_pi GT 1000) then mld_dbm_pi else 0

yes? let mask=if (mld_1000 NE 0) then 1

yes? let sigZ=sigma0[d=5,z=@ave]

yes? let sigma=sigZ*mask

yes? let sigT=sigma[l=@ave]

yes? list sigT

**ERROR Ferret crash; signal = 11

Any help would be appreciated. Thank you very much

Sarah