WARNS in /var/log/rspamd/rspamd.log

[Emanuel Gonzalez]

Hello,

I see two warns in the rspamd logs:

rspamd_worker_guard_handler: workaround for shutdown enabled, please update your client, this support might be removed in future

2018-05-24 11:16:29 #20195(normal) <adf61f>; lua; neural.lua:492: cannot learn ANN tRFANND41F6CDA875D14DB260: too many ham samples: 40

I disable the autolearn in the statistic.conf:

classifier "bayes" {
    tokenizer {
    name = "osb";
    }

    backend = "redis";
    servers = "127.0.0.1:6379";
    db = 3;
    #min_tokens = 11;
    #min_learns = 200;
    #autolearn = [-5, 15];
    #autolearn = true;

    per_user = <<EOD
return function(task)
    local rcpt = task:get_recipients(1)

if rcpt then
    one_rcpt = rcpt[1]
    if one_rcpt['domain'] then
        return one_rcpt['domain']
    end
end

return nil
end
EOD

    statfile {
        symbol = "BAYES_HAM";
        spam = false;
    }
    statfile {
        symbol = "BAYES_SPAM";
        spam = true;
    }
    learn_condition =<<EOD
return function(task, is_spam, is_unlearn)
    local prob = task:get_mempool():get_variable('bayes_prob', 'double')

    if prob then
        local in_class = false
        local cl
        if is_spam then
            cl = 'spam'
            in_class = prob >= 0.95
        else
            cl = 'ham'
            in_class = prob <= 0.05
        end

        if in_class then
            return false,string.format('already in class %s; probability %.2f%%',
            cl, math.abs((prob - 0.5) * 200.0))
        end
    end

    return true
end
EOD
}

the service is working very well but I want to know if these warnings can be harmful to my system

regards

[Alexander Moisseev]

On 24.05.2018 17:19, Emanuel Gonzalez wrote:
>
> 2018-05-24 11:16:29 #20195(normal) <adf61f>; lua; neural.lua:492: cannot learn ANN tRFANND41F6CDA875D14DB260: too many ham samples: 40
>

It is not a warning. Neural module skips messages to keep ham/spam learns balance. It is quite normal.

> I disable the autolearn in the statistic.conf:
>

You have disabled autolearning for Bayesian statistics. It is not related to neural module.

[Emanuel Gonzalez]

it is convenient to deactivate the module?

--
You received this message because you are subscribed to the Google Groups "rspamd" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rspamd+unsubscribe@googlegroups.com.
Visit this group at https://groups.google.com/group/rspamd.

[Alexander Moisseev]

On 24.05.2018 19:14, Emanuel Gonzalez wrote:
> it is convenient to deactivate the module?
>

> 2018-05-24 12:49 GMT-03:00 Alexander Moisseev <alexander...@gmail.com <mailto:alexander...@gmail.com>>:

>
> On 24.05.2018 17:19, Emanuel Gonzalez wrote:
>
>
> 2018-05-24 11:16:29 #20195(normal) <adf61f>; lua; neural.lua:492: cannot learn ANN tRFANND41F6CDA875D14DB260: too many ham samples: 40
>
> It is not a warning. Neural module skips messages to keep ham/spam learns balance. It is quite normal.
>
> I disable the autolearn in the statistic.conf:
>
> You have disabled autolearning for Bayesian statistics. It is not related to neural module.
>

It is up to you to decide if you need this module.
Neural network module is optional and disabled by default.
Take in consideration amount of mail traffic. If it is too low, neural network is useless as it never finish learning.

[Emanuel Gonzalez]

at the moment it is in test mode, but when it is in production it would be analyzing 30000 emails per hour

2018-05-24 13:47 GMT-03:00 Alexander Moisseev <alexander...@gmail.com>:

On 24.05.2018 19:14, Emanuel Gonzalez wrote:

it is convenient to deactivate the module?

2018-05-24 12:49 GMT-03:00 Alexander Moisseev <alexander...@gmail.com <mailto:alexander.moisseev@gmail.com>>:

    On 24.05.2018 17:19, Emanuel Gonzalez wrote:


        2018-05-24 11:16:29 #20195(normal) <adf61f>; lua; neural.lua:492: cannot learn ANN tRFANND41F6CDA875D14DB260: too many ham samples: 40

    It is not a warning. Neural module skips messages to keep ham/spam learns balance. It is quite normal.

        I disable the autolearn in the statistic.conf:

    You have disabled autolearning for Bayesian statistics. It is not related to neural module.

[Emanuel Gonzalez]

this warning referred to?

rspamd_worker_guard_handler: workaround for shutdown enabled, please update your client, this support might be removed in future

[Emanuel Gonzalez]

this warning: rspamd_worker_guard_handler: workaround for shutdown enabled, please update your client, this support might be removed in future

referred to?

2018-05-24 14:07 GMT-03:00 Emanuel Gonzalez <ema...@gmail.com>: