[developers] memory issues in [incr tsdb()]

[Rebecca Dridan]

Thanks Dan,

I eventually found that bit of the wiki last week, and switched to the 
bclim image. It does help, though I still run out of memory if I try 
updating the full set of 400-odd trees. Dividing the set in half now 
works though.  I'm not sure how adjustable that image is - maybe some 
Lisp experts out there could let me know if it's safe to increase some 
of the numbers?  I guess I'll see some of you in Berlin next week anyway 
to ask.

Thanks,

Rebecca

Dan Flickinger wrote:
> Hi Rebecca -
>
> If you're using a vanilla Lisp image, it may not allow a large enough
> heap for what you want to do.  You (or the system administrator if 
> you're not running on your own machine) can build a customized Lisp 
> image which allows a much larger heap.  See the file 
>  $DELPHINHOME/lkb/etc/bclim.lisp
> (which is also pointed to on the DELPH-IN wiki, on the LKBCompilation
> page).
>
> Once you've built this image, which will be called "bclim.dxl", you'll 
> need to adjust your local settings so you invoke this image when you
> start up your Lisp.
>
> If this is too cryptic, maybe you can say more about your current
> set-up (operating system, machine, etc.), and we can try to be more
> precise about the steps needed.
>
>  Dan
>
>   
>> Hi,
>>
>> I keep running into:
>>
>> Error: An allocation request for 16 bytes caused a need for 122159104 more
>>        bytes of heap.  The operating system will not make the space 
>> available
>>        because of a lack of swap space or some other operating system 
>> imposed
>>        limit or memory mapping collision.
>>   [condition type: STORAGE-CONDITION]
>>
>> whenever I try and do an automatic update of a tree in [incr tsdb()] on 
>> a 32-bit system. I thought it was because I was running out of memory on 
>> my system and so switched to a different system but it still happens 
>> when I have plenty of memory, I'm not even hitting swap. Incidently, it 
>> doesn't happen when I switch to a 64-bit machine, but that causes other 
>> issues. I've trawled the archives looking for solutions but maybe I've 
>> missed something? Is there any variable (environment, [incr tsdb()], 
>> etc) that I can set to stop it crashing?
>>
>> Thanks,
>>
>> Rebecca
>>     
>
>