[Gross] false positives

[Eino Tuominen]

Imri Zvik wrote:
> On Wednesday 17 December 2008 19:16:07 Eino Tuominen wrote:
>> Imri Zvik wrote:
>>> I forgot to CC the list on my reply to Jesse:
>>>
>>> If I raise the filter bit, grossd crashes with segmentation fault.
>> Please, show your configuration file that triggered the crash. At least
>> filter_bits, number_buffers and rotate_interval.
> 
> 
> filter_bits = 26
> number_buffers=32
> rotate_interval=10800
> 
> If I raise the filter_bits anywhere above 26, it would crash after a few 
> seconds (while replicating/syncing with the other peer).

With your settings you end up using over 2 gigabytes of memory - if you 
increase filter_bits even by one it will use more than 4 GB. Which is 
more than a 32 bit virtual memory can map.

I suggest you start using block_treshold feature. If that's really not 
an option, try something like this:

filter_bits = 28
number_buffers = 8
rotate_interval = 43200

Also, your expire time is a bit long, 24 hours should be enough always.

filter_bits = 28
number_buffers = 9
rotate_interval = 10800

You can calculate the needed memory in bytes with this formula:

2^filter_bits * ( number_buffers + 1 )

Filter expiration time is (in seconds) number_buffers * rotate interval

Hope this helps,

-- 
   Eino Tuominen