[Rspamd-Users] Questions about REDIS and Bayes Learning

[Dan Swartzendruber]

On 2018-11-19 15:54, Alexander Moisseev via Users wrote:
> On 19.11.18 23:33, Dan Swartzendruber wrote:
>> On 2018-11-19 15:26, Alexander Moisseev via Users wrote:
>>> On 19.11.18 22:49, Dan Swartzendruber wrote:
>>>> On 2018-11-19 14:20, Alexander Moisseev via Users wrote:
>>>>> 
>>>>> Are you sure you are using "redis" backend?
>>>>> 
>>>>> # rspamadm configdump classifier.bayes.backend
>>>> 
>>>> Crap, I'm not :(
>>>> 
>>>> *** Section classifier.bayes.backend ***
>>>> "sqlite3"
>>>> *** End of section classifier.bayes.backend ***
>>>> 
>>>> I think I was just using whatever defaults there are.  Is it okay to 
>>>> leave as is?
>>>> 
>>>> 
>>> 
>>> Redis backend and "new schema" are highly recommended.
>>> I would suggest to use a separate Redis instance for Bayes 
>>> statistics.
>>> It makes life easier.
>>> For example, you can just copy .rdb file to make backup, share it
>>> between installation or limit memory usage like that:
>>> https://rspamd.com/doc/modules/bayes_expiry.html#limiting-memory-usage-to-a-fixed-amount
>> 
>> Is there a way to migrate the sqlite DB, or is that not needed?
> 
> `rspamadm configwizard statistics` or you can just start it over as
> you have learned only about 150 messages.

okay, thanks!