Topic: Spam filter question

[Amir Inbar]

Does anybody know a way to enable or dissable or change settings of the spam filter on SME7rc2 per user and globally ?
Is it possible at all ?

I would like to have the option to set strong filtering for specific mailboxes and light filtering for other mailboxes on the same server.

Amir

[idyll]

The spamassassin installation is global. You would need to have discreet installations for each user, in each user's home path for the granular control you seek.

You might consider client-side spam filtering. Thunderbird, and others, provide Bayes filtering per client and it works pretty well.

regards,

patrick

[Amir Inbar]

Thank you idyll for the answer.
I know the client side solution but this is exactly what i want to avoid - i am seeking a centralized control over the network and since different neworks has different users and different user-security-habbits i think that a way to set filtering depth to different users is a must.

Convincing users to change habbits is almost impossible (as you probably know) and hopping they will start to use Thunderbird instead of OE or MO is not something i would consider as an option even though i install TB to most of my users computers on most of my controlled networks  ...

I just wondered if there is a way to do it on SME7rc2 and Spamassassin - you said "discreet installations" - what exacly do you mean ?

[idyll]

I proposed a hopelessly complex task, and one I do not recommend. I mention it only for context.

discreet installation = one operating instance of spamassassin per user. NO ONE does this with any scaled operation. But I have seen it done for five or six users.

I understand what you are seeking, and I personally have no knowledge of a simple way to do it.

Perhaps a moderately aggressive filtering control, with a larger than normal number of "tagged" messages - plus large scale whitelisting is a solution? You could solicit whitelist addresses before the fact, and generate a reasonably wide-spread whitelist at cut-over. With the large "tagged" message base you could request users submit addresses, etc. as mis-tagged messages accumulate in their junkmail folders.

I sympathize wit you, but I found the above scenario becomes manageable in a reasonably short time and virtually nothing is deleted which isn't pure spam.

regards,

patrick

[Amir Inbar]

Patrick thank you for the idea but i have some problems with that and correct me if i miss-understood you :

You propose creating a whitelist to be used as positive filter for all the mails while raising the negative filter to a very high level.

1. this will bind me or the admin to deal with users complains and requests - constantly changing the whitelist when they disscover something was filtered by mistake.

2. Creating the basic whitelist will consume a very large amount of time for each location.

3. this will block unknown new senders.

4. i will have to get a focus on what are the rules that set the flag of spam to all the messages - for example - if a spam rule at the higher levels of filtering is taging a message as spam just because the sender sent it to more than 40 recipients - it could be a problem...

Thank you for the creative thinking anyhow...

[pfloor]

Amir,

The stock settings in the server manager are as close to perfect as you can get.  Here is my spam over the last 6 weeks:

898 spam detected with 3 that were not and 3 additional that got through.

Spam control will never be perfect but this is 99.6% and you can't ask for anythig better than that.