You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by "Michael Hallager (personal)" <mi...@nettrust.nz> on 2018/07/17 03:53:34 UTC
SA MySQL DB maintenance
Hi all,
Does SA self-maintain the records in the DB or is there a script I
should run periodically for this?
Michael
Re: SA MySQL DB maintenance
Posted by Axb <ax...@gmail.com>.
On 07/17/2018 08:29 PM, Kevin A. McGrail wrote:
> I'll defer that question to Alex Broens. He can do it more justice than I
> ever could. AXB?
>
> --
> Kevin A. McGrail
> VP Fundraising, Apache Software Foundation
> Chair Emeritus Apache SpamAssassin Project
> https://www.linkedin.com/in/kmcgrail - 703.798.0171
>
> On Tue, Jul 17, 2018 at 1:55 PM, micah anderson <mi...@riseup.net> wrote:
>
>> "Kevin A. McGrail" <km...@apache.org> writes:
>>
>>> I think Bayes should be in redis though not SQL.
>>
>> Curious to know why you think that?
+ amazing speed
+ set & forget / self maintenance
+ stability
- depending on your setup, it may require quite a lot of cheap RAM.
More details:
https://svn.apache.org/repos/asf/spamassassin/trunk/contrib/HOWTO.Bayes-Redis/
h2h
Axb
Re: SA MySQL DB maintenance
Posted by "Kevin A. McGrail" <km...@apache.org>.
I'll defer that question to Alex Broens. He can do it more justice than I
ever could. AXB?
--
Kevin A. McGrail
VP Fundraising, Apache Software Foundation
Chair Emeritus Apache SpamAssassin Project
https://www.linkedin.com/in/kmcgrail - 703.798.0171
On Tue, Jul 17, 2018 at 1:55 PM, micah anderson <mi...@riseup.net> wrote:
> "Kevin A. McGrail" <km...@apache.org> writes:
>
> > I think Bayes should be in redis though not SQL.
>
> Curious to know why you think that?
>
Re: SA MySQL DB maintenance
Posted by micah anderson <mi...@riseup.net>.
"Kevin A. McGrail" <km...@apache.org> writes:
> I think Bayes should be in redis though not SQL.
Curious to know why you think that?
Re: SA MySQL DB maintenance
Posted by RW <rw...@googlemail.com>.
On Wed, 18 Jul 2018 10:49:16 +1200
Michael Hallager (personal) wrote:
> On 2018-07-18 01:11, Giovanni Bechis wrote:
> > Txrep does not have autocleaning support, bayes have it if
> > auto_bayes_expire is set.
> > Giovanni
>
> I have looked into this here:
> https://spamassassin.apache.org/full/3.1.x/doc/sa-learn.html#expiration
>
> Our Bayes DB is over 5Gb which seems at odds with the auto purge
> figures mentioned at the above link.
>
> Even running "sa-learn --auto-expire" does not impact the size.
There's no such thing as "sa-learn --auto-expire".
As I previously mentioned auto-expiry runs during a scan which can mean
that both the scan and auto-expiry are aborted by a time-out.
Re: SA MySQL DB maintenance
Posted by "Michael Hallager (personal)" <mi...@nettrust.nz>.
On 2018-07-18 01:11, Giovanni Bechis wrote:
> Txrep does not have autocleaning support, bayes have it if
> auto_bayes_expire is set.
> Giovanni
I have looked into this here:
https://spamassassin.apache.org/full/3.1.x/doc/sa-learn.html#expiration
Our Bayes DB is over 5Gb which seems at odds with the auto purge figures
mentioned at the above link.
Even running "sa-learn --auto-expire" does not impact the size.
------
Further to my above post, running "sa-learn --auto-expire" with
"bayes_auto_expire 0" set in the config produces a different result,
namely the database is now down to a few 10's of Mb.
I am taking the advice provided here and running this command as a
cronjob.
Many thanks.
Michael
Re: SA MySQL DB maintenance
Posted by Benny Pedersen <me...@junc.eu>.
Kevin A. McGrail skrev den 2018-07-17 18:40:
> RW is right. Don't use auto_bayes_expire use sa-learn --force-expire
> in a cron. It's another bayes option I've wondered "why does this
> exist"?
it was usefull to keep all bayes digest before ?
my bayes_seen is bigger then my bayes_digest
auto_bayes_expire 0
will keep it as learnt ?
who will add Mail::DKIM::ARC to spamassassin plugin ?
already done in trunk ?
needs Mail::DKIM 0.440.0
Re: SA MySQL DB maintenance
Posted by "Kevin A. McGrail" <km...@apache.org>.
RW is right. Don't use auto_bayes_expire use sa-learn --force-expire in a
cron. It's another bayes option I've wondered "why does this exist"?
--
Kevin A. McGrail
VP Fundraising, Apache Software Foundation
Chair Emeritus Apache SpamAssassin Project
https://www.linkedin.com/in/kmcgrail - 703.798.0171
On Tue, Jul 17, 2018 at 11:54 AM, RW <rw...@googlemail.com> wrote:
> On Tue, 17 Jul 2018 15:11:10 +0200
> Giovanni Bechis wrote:
>
> > Txrep does not have autocleaning support, bayes have it
>
> It's not complete as there is no way of ageing-out the entries that
> record whether a particular email was trained as spam or ham.
>
> > if auto_bayes_expire is set.
>
> It's safer to turn that off and use sa-learn --force-expire. autoexpiry
> runs as part of a scan and can cause timeouts.
>
Re: SA MySQL DB maintenance
Posted by RW <rw...@googlemail.com>.
On Tue, 17 Jul 2018 15:11:10 +0200
Giovanni Bechis wrote:
> Txrep does not have autocleaning support, bayes have it
It's not complete as there is no way of ageing-out the entries that
record whether a particular email was trained as spam or ham.
> if auto_bayes_expire is set.
It's safer to turn that off and use sa-learn --force-expire. autoexpiry
runs as part of a scan and can cause timeouts.
Re: SA MySQL DB maintenance
Posted by "Michael Hallager (personal)" <mi...@nettrust.nz>.
On 2018-07-18 01:11, Giovanni Bechis wrote:
> Txrep does not have autocleaning support, bayes have it if
> auto_bayes_expire is set.
> Giovanni
Hi Giovanni,
I have looked into this here:
https://spamassassin.apache.org/full/3.1.x/doc/sa-learn.html#expiration
Our Bayes DB is over 5Gb which seems at odds with the auto purge figures
mentioned at the above link.
Even running "sa-learn --auto-expire" does not impact the size.
Michael
Re: SA MySQL DB maintenance
Posted by Giovanni Bechis <gi...@paclan.it>.
Txrep does not have autocleaning support, bayes have it if auto_bayes_expire is set.
Giovanni
On 07/17/18 14:35, Kevin A. McGrail wrote:
> To me, no, it doesn't.
>
> For example, I clean out txrep stuff with crons like this -e 'DELETE FROM txrep WHERE last_hit <= (now() - INTERVAL 90 day);'
>
> I also don't use autolearning bayes but some people like to cull there bayes. I think Bayes should be in redis though not SQL.
>
> What are you using in a db?
>
> Regards,
> KAM
>
> --
> Kevin A. McGrail
> VP Fundraising, Apache Software Foundation
> Chair Emeritus Apache SpamAssassin Project
> https://www.linkedin.com/in/kmcgrail - 703.798.0171
>
> On Mon, Jul 16, 2018 at 11:53 PM, Michael Hallager (personal) <michael@nettrust.nz <ma...@nettrust.nz>> wrote:
>
> Hi all,
>
> Does SA self-maintain the records in the DB or is there a script I should run periodically for this?
>
> Michael
>
>
Re: SA MySQL DB maintenance
Posted by "Michael Hallager (personal)" <mi...@nettrust.nz>.
On 2018-07-18 00:35, Kevin A. McGrail wrote:
> What are you using in a db?
>
> Regards,
> KAM
>
> --
> Kevin A. McGrail
Hi Kevin,
awl, bayes_* and userpref.
Thanks,
Michael
Re: SA MySQL DB maintenance
Posted by "Kevin A. McGrail" <km...@apache.org>.
To me, no, it doesn't.
For example, I clean out txrep stuff with crons like this -e 'DELETE FROM
txrep WHERE last_hit <= (now() - INTERVAL 90 day);'
I also don't use autolearning bayes but some people like to cull there
bayes. I think Bayes should be in redis though not SQL.
What are you using in a db?
Regards,
KAM
--
Kevin A. McGrail
VP Fundraising, Apache Software Foundation
Chair Emeritus Apache SpamAssassin Project
https://www.linkedin.com/in/kmcgrail - 703.798.0171
On Mon, Jul 16, 2018 at 11:53 PM, Michael Hallager (personal) <
michael@nettrust.nz> wrote:
> Hi all,
>
> Does SA self-maintain the records in the DB or is there a script I should
> run periodically for this?
>
> Michael
>