Re: [SAtalk] Bayes versioning?

2003-11-12 Thread Ted Cabeen
Theo Van Dinter <[EMAIL PROTECTED]> writes:

> On Tue, Nov 11, 2003 at 01:15:43PM -0800, Ted Cabeen wrote:
>> Okay.  I'll give that a try.  What's the best way to quiesce the bayes
>> system without making spamassassin unavailable during the import?
>> Will setting auto_learn to 0 in the running user's user_prefs be
>> enough to make the import go smoothly, or should I set use_bayes to 0
>> as well?
>
> auto_learn 0  will disable automatic writes (which will fail anyway),
> but use_bayes 0  will fully disable any access of the db.  I'd do that.

I just tried to run the import and the process ran out of memory.
I've got 1GB of physical RAM in that machine with 2GB of swap
available and it still errored out with the following output:
# sa-learn --dbpath /cyrus/amavis/.spamassassin/bayes --import
upgrading to DB_File, please be patient: /cyrus/amavis/.spamassassin/old_bayes_seen
DB_File: copied 702806 entries.
upgrading to DB_File, please be patient: /cyrus/amavis/.spamassassin/old_bayes_toks
DB_File: DB_File module not installed, nothing copied.
GDBM_File: GDBM_File module not installed, nothing copied.
Out of memory during request for 1012 bytes, total sbrk() is 1073326080 bytes!

I ran the import as root with all of the ulimits set to unlimited.  Is
there any way to get the memory usage down so that I can complete the
import?

Can I somehow purge the less important data from the database?

-- 
Ted Cabeen   http://www.pobox.com/~secabeen[EMAIL PROTECTED] 
Check Website or Keyserver for PGP/GPG Key BA0349D2 [EMAIL PROTECTED]
"I have taken all knowledge to be my province." -F. Bacon  [EMAIL PROTECTED]
"Human kind cannot bear very much reality."-T.S.Eliot[EMAIL PROTECTED]


---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Ted Cabeen
Rod Begbie <[EMAIL PROTECTED]> writes:

> -BEGIN PGP SIGNED MESSAGE-
> Hash: SHA1
>
> Ted Cabeen wrote:
>
>> I've got a site-wide bayes database that I'm not sure is correctly
>> configured.  It's quite a large database, and I think it may not have
>> been converted to the new DB format when I upgraded.  When I try to
>> run a sa-learn --dump magic, I get the following output:
>
> I'd use the --dbpath option to make sure you're pointing sa-learn to
> the right bayes_ files.
>
> My guess is it's looking in the wrong place.

Close.  It wasn't looking in the wrong place but I was running
sa-learn as the wrong user.  Duh.  However, when I run it correctly, I
get the following output, which clearly shows that I'm running DB
version 0:
0.000  0  0  0  non-token data: bayes db version
0.000  0 101949  0  non-token data: nspam
0.000  0  26120  0  non-token data: nham
0.000  02174480  0  non-token data: ntokens
0.000  0  0  0  non-token data: oldest atime
0.000  0  52652  0  non-token data: current scan-count
0.000  0  0  0  non-token data: last expiry atime

That said, I'm going to try running sa-learn --import again to try to
go to db version 2 once I figure out the best way to quiesce the bayes
system without disabling SA entirely.

Thanks for the reminder.

-- 
Ted Cabeen   http://www.pobox.com/~secabeen[EMAIL PROTECTED] 
Check Website or Keyserver for PGP/GPG Key BA0349D2 [EMAIL PROTECTED]
"I have taken all knowledge to be my province." -F. Bacon  [EMAIL PROTECTED]
"Human kind cannot bear very much reality."-T.S.Eliot[EMAIL PROTECTED]


---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Ted Cabeen
Theo Van Dinter <[EMAIL PROTECTED]> writes:

> On Tue, Nov 11, 2003 at 01:00:05PM -0800, Ted Cabeen wrote:
>> I was running perl 5.005 back then, but I've upgraded to 5.8.1 now.
>> Should I try running the import again? 
>
> You might want to try that.  The error you're reporting seems to occur
> when the DB is not accessible, likely for writing.  Since you still have
> the msgcount file, the DB has not been updated from v0 to v2.

Okay.  I'll give that a try.  What's the best way to quiesce the bayes
system without making spamassassin unavailable during the import?
Will setting auto_learn to 0 in the running user's user_prefs be
enough to make the import go smoothly, or should I set use_bayes to 0
as well?

-- 
Ted Cabeen   http://www.pobox.com/~secabeen[EMAIL PROTECTED] 
Check Website or Keyserver for PGP/GPG Key BA0349D2 [EMAIL PROTECTED]
"I have taken all knowledge to be my province." -F. Bacon  [EMAIL PROTECTED]
"Human kind cannot bear very much reality."-T.S.Eliot[EMAIL PROTECTED]


---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Rod Begbie
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Ted Cabeen wrote:

I've got a site-wide bayes database that I'm not sure is correctly
configured.  It's quite a large database, and I think it may not have
been converted to the new DB format when I upgraded.  When I try to
run a sa-learn --dump magic, I get the following output:
I'd use the --dbpath option to make sure you're pointing sa-learn to the 
right bayes_ files.

My guess is it's looking in the wrong place.

Rod.

- -- 
:: Rod Begbie :: http://www.groovymother.com/ ::

C is for Cookie.  That's good enough for me.
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.2.3-nr1 (Windows XP)
iD8DBQE/sVONN/YdXIyAw18RAq8SAKDdiIATJTL0lQVkLOMx91B9eO2RsgCfeQ6N
OSpikahK6FraNr+75NPsbj8=
=dP3s
-END PGP SIGNATURE-
---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Theo Van Dinter
On Tue, Nov 11, 2003 at 01:00:05PM -0800, Ted Cabeen wrote:
> I was running perl 5.005 back then, but I've upgraded to 5.8.1 now.
> Should I try running the import again? 

You might want to try that.  The error you're reporting seems to occur
when the DB is not accessible, likely for writing.  Since you still have
the msgcount file, the DB has not been updated from v0 to v2.

-- 
Randomly Generated Tagline:
"Don't ever make trouble here, I beat you up each time."
  - From Rumble in the Bronx


pgp0.pgp
Description: PGP signature


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Ted Cabeen
Theo Van Dinter <[EMAIL PROTECTED]> writes:

> On Tue, Nov 11, 2003 at 12:36:54PM -0800, Ted Cabeen wrote:
>> Here are the directory entries:
>> -rw---  1 amavisd  amavisd  3148 Nov 11 12:29 bayes_journal
>> -rw---  1 amavisd  amavisd  2652 Oct  1 10:36 bayes_msgcount
>> -rw---  1 amavisd  amavisd  41607168 Nov 11 12:29 bayes_seen
>> -rw---  1 amavisd  amavisd  85565440 Nov 11 12:29 bayes_toks
>> 
>> The database is working, as I do see BAYES_99 rules being triggered in
>> my inbound spams, but I'm concerned about the database format and
>> about losing all of the data it's already built up over the last year.
>
> Did you read the docs, make sure you have DB_File installed, and run
> "sa-learn --import"?

I did.  I recall it took at least 4 hours and over 500MB of RAM to run
the import, and I wasn't positive that it actually went through.
DB_File was definitely installed (it comes with FBSD perl).

I was running perl 5.005 back then, but I've upgraded to 5.8.1 now.
Should I try running the import again? 

-- 
Ted Cabeen   http://www.pobox.com/~secabeen[EMAIL PROTECTED] 
Check Website or Keyserver for PGP/GPG Key BA0349D2 [EMAIL PROTECTED]
"I have taken all knowledge to be my province." -F. Bacon  [EMAIL PROTECTED]
"Human kind cannot bear very much reality."-T.S.Eliot[EMAIL PROTECTED]


---
This SF.Net email sponsored by: ApacheCon 2003,
16-19 November in Las Vegas. Learn firsthand the latest
developments in Apache, PHP, Perl, XML, Java, MySQL,
WebDAV, and more! http://www.apachecon.com/
___
Spamassassin-talk mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/spamassassin-talk


Re: [SAtalk] Bayes versioning?

2003-11-11 Thread Theo Van Dinter
On Tue, Nov 11, 2003 at 12:36:54PM -0800, Ted Cabeen wrote:
> Here are the directory entries:
> -rw---  1 amavisd  amavisd  3148 Nov 11 12:29 bayes_journal
> -rw---  1 amavisd  amavisd  2652 Oct  1 10:36 bayes_msgcount
> -rw---  1 amavisd  amavisd  41607168 Nov 11 12:29 bayes_seen
> -rw---  1 amavisd  amavisd  85565440 Nov 11 12:29 bayes_toks
> 
> The database is working, as I do see BAYES_99 rules being triggered in
> my inbound spams, but I'm concerned about the database format and
> about losing all of the data it's already built up over the last year.

Did you read the docs, make sure you have DB_File installed, and run
"sa-learn --import"?

-- 
Randomly Generated Tagline:
"...you might as well skip the Xmas celebration completely, and instead
 sit in front of your linux computer playing with the
 all-new-and-improved linux kernel version."
 (By Linus Torvalds)


pgp0.pgp
Description: PGP signature