You are viewing a plain text version of this content. The canonical link for it is here.
Posted to ruleqa@spamassassin.apache.org by Kent Oyer <ke...@mxguardian.net> on 2026/02/14 00:23:55 UTC
Re[2]: RuleQA low on fuel…
I'm also a little confused about the masscheck run time. The docs say "at or after 9AM UTC". But since this is a daily job, aren't all times technically after 9AM? (i.e. 6AM is after 9AM the day before). Perhaps it would be better to specify a time range. Job needs to start after X but be finished before Y.
On Fri, Feb 13, 2026 at 11:24 AM, Lucian Langa <lu...@gnome.eu.org> wrote:
------------------------------------------------------------------------------------------------------------------
CAUTION: External email from: lucilanga@gnome.eu.org (mailto:lucilanga@gnome.eu.org)
Use caution before clicking on links or opening attachments.
------------------------------------------------------------------------------------------------------------------
Hi Bill,
On 13.02.2026 10:57, Bill Cole wrote:
Anyone have ideas for how to fix this?
could this once again be that (my) submission time is off (though I
might not be the only one)
Looking at checkMasscheckContribs.sh time is 15:17 UTC
SVN tagged rev in nightly_mass_check: 1931846
New masscheck submission listings in the past day:
SVN rev (Match) File Name (Date)
1931824 (No) - ham-llanga.r1931824.log (Feb 12 17:26)
but rev 1931846 was completed on my side at about Feb 13 18:34 CET
(that's 17:34 UTC) so it probably means is was late at least 2 hours.
yet on "in the past week" rev 1031743 seems fine at 17:25 (10 mins
earlier than 13'th run.
Recent masscheck submission listings in the past week:
SVN rev (Match) File Name (Date)
1931743 (Yes) - ham-net-llanga.r1931743.log (Feb 7 17:25)
I must admit never know exactly the time should run my script and to be
perfectly honest ".. start your run after H hour" doesn't tell a much.
The corpus is variable, the machine running the corpus also a variable
and the day of the week (Sat is slower because of the -net rules).
It would be much clear (at least for me) if I know the time when I have
to rsync my corpus - say make sure your corpus hits the server by H
hour. That way I can adapt my starting script time with the corpus size,
machine and the day of the week.
See the message below. Also related errors reported to both the sysadmins and ruleqa lists for over a week.
I only got "bad performing rules" mail that doesn't tell much about
submission.
Can we (ruleqa) get the output of this script too and there
was another (I can't remember its name) about the corpus of each
contributor and the minimum threshold, that could be also useful.
--
Cheers,
--lucian
Re: RuleQA low on fuel…
Posted by Bill Cole <sa...@billmail.scconsult.com>.
On 2026-02-14 at 14:26:14 UTC-0500 (Sat, 14 Feb 2026 19:26:14 +0000)
Kent Oyer <ru...@spamassassin.apache.org>
is rumored to have said:
> Thanks for clarifying the start & end times. What does no match mean:
> SVN rev (Match) File Name (Date) 1931743 (No) -
> ham-net-koyer.r1931743.log (Feb 14 06:04)
I have not analyzed the script but my deduction from the output is that
"Match" tells whether the SVN rev used for the masscheck matches the
correct version tagged for the rescoring run.
Re: RuleQA low on fuel…
Posted by Bill Cole <sa...@billmail.scconsult.com>.
On 2026-02-14 at 14:26:14 UTC-0500 (Sat, 14 Feb 2026 19:26:14 +0000)
Kent Oyer <ru...@spamassassin.apache.org>
is rumored to have said:
> Thanks for clarifying the start & end times. What does no match mean:
> SVN rev (Match) File Name (Date)
> 1931743 (No) - ham-net-koyer.r1931743.log (Feb 14 06:04)
It means that the SVN revision of the rules used for the masscheck isn't
the one that the server wants for its rescore.
--
Bill Cole
bill@scconsult.com or billcole@apache.org
(AKA @grumpybozo@toad.social and many *@billmail.scconsult.com
addresses)
Please keep discussion mailing list replies *on-list*
Not Currently Available For Hire
Re[2]: RuleQA low on fuel…
Posted by Kent Oyer <ke...@mxguardian.net>.
Thanks for clarifying the start & end times. What does no match mean:
SVN rev (Match) File Name (Date) 1931743 (No) - ham-net-koyer.r1931743.log (Feb 14 06:04)
On Sat, Feb 14, 2026 at 09:13 AM, Bill Cole <sa...@billmail.scconsult.com> wrote:
NOTE: All times below referring to sa-vm ops are UTC.
On 2026-02-13 at 19:23:55 UTC-0500 (Sat, 14 Feb 2026 00:23:55 +0000)
Kent Oyer <ruleqa@spamassassin.apache.org (mailto:ruleqa@spamassassin.apache.org)>
is rumored to have said:
I'm also a little confused about the masscheck run time. The docs say
"at or after 9AM UTC". But since this is a daily job, aren't all times
technically after 9AM? (i.e. 6AM is after 9AM the day before).
I think the vagueness is due to everything involved taking a long
time...
"Daily at some time after 0900 UTC" means after 0900 UTC every day. Not
after yesterday's 0900.
Perhaps it would be better to specify a time range. Job needs to start
after X but be finished before Y.
OK, here's an arbitrary unilateral spec: after 0900 UTC, finish by 2300
UTC.
I believe that should work for all cases. Not sure how long the job is
taking for anyone.
On Fri, Feb 13, 2026 at 11:24 AM, Lucian Langa
<lucilanga@gnome.eu.org (mailto:lucilanga@gnome.eu.org)> wrote:
Hi Bill,
On 13.02.2026 10:57, Bill Cole wrote:
Anyone have ideas for how to fix this?
could this once again be that (my) submission time is off (though I
might not be the only one)
Looking at checkMasscheckContribs.sh time is 15:17 UTC
That script runs hourly.
SVN tagged rev in nightly_mass_check: 1931846
New masscheck submission listings in the past day:
SVN rev (Match) File Name (Date)
1931824 (No) - ham-llanga.r1931824.log (Feb 12 17:26)
but rev 1931846 was completed on my side at about Feb 13 18:34 CET
(that's 17:34 UTC) so it probably means is was late at least 2 hours.
yet on "in the past week" rev 1031743 seems fine at 17:25 (10 mins
earlier than 13'th run.
Confession: I am NOT expert in how the masscheck/ruleqa system is
scheduled or precisely which cron jobs do what. I just noticed the
chronic failures, coming to sysadmins@spamassassin.apache.org (mailto:sysadmins@spamassassin.apache.org).
I mean, obviously I can read the automc crontab, but I have not read all
the scripts and don't have a flow diagram in my head. Some scripts tell
lies in their comments...
Recent masscheck submission listings in the past week:
SVN rev (Match) File Name (Date)
1931743 (Yes) - ham-net-llanga.r1931743.log (Feb 7 17:25)
I must admit never know exactly the time should run my script and to
be
perfectly honest ".. start your run after H hour" doesn't tell a much.
The corpus is variable, the machine running the corpus also a variable
and the day of the week (Sat is slower because of the -net rules).
It would be much clear (at least for me) if I know the time when I
have
to rsync my corpus - say make sure your corpus hits the server by H
hour. That way I can adapt my starting script time with the corpus
size,
machine and the day of the week.
Based on the crontab, which has once-daily jobs starting 01:20-08:37,
along with the existing documentation not giving a deadline, I think a
safe deadline is 23:00.
See the message below. Also related errors reported to both the
sysadmins and ruleqa lists for over a week.
I only got "bad performing rules" mail that doesn't tell much about
submission.
All mail from cron jobs in the automc crontab goes to the sysadmins
list. The nightly rescore (do-stable-update-with-scores) runs some
scripts whose ouput is delivered to ruleqa@spamassassin.apache.org (mailto:ruleqa@spamassassin.apache.org),
notably with subjects like "[auto] do-nightly-rescore-example return
code 9" where the code varies. They come from "automc Cron
<noreply@spamassassin.apache.org (mailto:noreply@spamassassin.apache.org)>" so it is not a huge leap of
imagination to think that it could be caught by filters. The envelope
sender is a per-user VERP address.
Can we (ruleqa) get the output of this script too and there
was another (I can't remember its name) about the corpus of each
contributor and the minimum threshold, that could be also useful.
See https://lists.apache.org/list?ruleqa@spamassassin.apache.org:2026-2 (https://lists.apache.org/list?ruleqa@spamassassin.apache.org:2026-2)
for everything sent to the ruleqa list this month and check to make sure
that you are not blocking the messages. See
https://lists.apache.org/list?sysadmins@spamassassin.apache.org:2026-2 (https://lists.apache.org/list?sysadmins@spamassassin.apache.org:2026-2)
for what is going to the sysadmins list. You may want to subscribe to
both lists. If you can identify something you think needs to be
otherwise directed, we can fix it. However, I think we are *trying* to
send what you want.
--
Bill Cole
bill@scconsult.com (mailto:bill@scconsult.com) or billcole@apache.org (mailto:billcole@apache.org)
(AKA @grumpybozo@toad.social (mailto:grumpybozo@toad.social) and many *@billmail.scconsult.com
addresses)
Not Currently Available For Hire
Re: RuleQA low on fuel…
Posted by Bill Cole <sa...@billmail.scconsult.com>.
NOTE: All times below referring to sa-vm ops are UTC.
On 2026-02-13 at 19:23:55 UTC-0500 (Sat, 14 Feb 2026 00:23:55 +0000)
Kent Oyer <ru...@spamassassin.apache.org>
is rumored to have said:
> I'm also a little confused about the masscheck run time. The docs say
> "at or after 9AM UTC". But since this is a daily job, aren't all times
> technically after 9AM? (i.e. 6AM is after 9AM the day before).
I think the vagueness is due to everything involved taking a long
time...
"Daily at some time after 0900 UTC" means after 0900 UTC every day. Not
after yesterday's 0900.
> Perhaps it would be better to specify a time range. Job needs to start
> after X but be finished before Y.
OK, here's an arbitrary unilateral spec: after 0900 UTC, finish by 2300
UTC.
I believe that should work for all cases. Not sure how long the job is
taking for anyone.
> On Fri, Feb 13, 2026 at 11:24 AM, Lucian Langa
> <lu...@gnome.eu.org> wrote:
> Hi Bill,
>
> On 13.02.2026 10:57, Bill Cole wrote:
> Anyone have ideas for how to fix this?
>
> could this once again be that (my) submission time is off (though I
> might not be the only one)
>
> Looking at checkMasscheckContribs.sh time is 15:17 UTC
That script runs hourly.
> SVN tagged rev in nightly_mass_check: 1931846
>
> New masscheck submission listings in the past day:
> SVN rev (Match) File Name (Date)
> 1931824 (No) - ham-llanga.r1931824.log (Feb 12 17:26)
>
> but rev 1931846 was completed on my side at about Feb 13 18:34 CET
> (that's 17:34 UTC) so it probably means is was late at least 2 hours.
>
> yet on "in the past week" rev 1031743 seems fine at 17:25 (10 mins
> earlier than 13'th run.
Confession: I am NOT expert in how the masscheck/ruleqa system is
scheduled or precisely which cron jobs do what. I just noticed the
chronic failures, coming to sysadmins@spamassassin.apache.org.
I mean, obviously I can read the automc crontab, but I have not read all
the scripts and don't have a flow diagram in my head. Some scripts tell
lies in their comments...
> Recent masscheck submission listings in the past week:
> SVN rev (Match) File Name (Date)
> 1931743 (Yes) - ham-net-llanga.r1931743.log (Feb 7 17:25)
>
> I must admit never know exactly the time should run my script and to
> be
> perfectly honest ".. start your run after H hour" doesn't tell a much.
> The corpus is variable, the machine running the corpus also a variable
> and the day of the week (Sat is slower because of the -net rules).
>
> It would be much clear (at least for me) if I know the time when I
> have
> to rsync my corpus - say make sure your corpus hits the server by H
> hour. That way I can adapt my starting script time with the corpus
> size,
> machine and the day of the week.
Based on the crontab, which has once-daily jobs starting 01:20-08:37,
along with the existing documentation not giving a deadline, I think a
safe deadline is 23:00.
> See the message below. Also related errors reported to both the
> sysadmins and ruleqa lists for over a week.
>
> I only got "bad performing rules" mail that doesn't tell much about
> submission.
All mail from cron jobs in the automc crontab goes to the sysadmins
list. The nightly rescore (do-stable-update-with-scores) runs some
scripts whose ouput is delivered to ruleqa@spamassassin.apache.org,
notably with subjects like "[auto] do-nightly-rescore-example return
code 9" where the code varies. They come from "automc Cron
<no...@spamassassin.apache.org>" so it is not a huge leap of
imagination to think that it could be caught by filters. The envelope
sender is a per-user VERP address.
> Can we (ruleqa) get the output of this script too and there
> was another (I can't remember its name) about the corpus of each
> contributor and the minimum threshold, that could be also useful.
See https://lists.apache.org/list?ruleqa@spamassassin.apache.org:2026-2
for everything sent to the ruleqa list this month and check to make sure
that you are not blocking the messages. See
https://lists.apache.org/list?sysadmins@spamassassin.apache.org:2026-2
for what is going to the sysadmins list. You may want to subscribe to
both lists. If you can identify something you think needs to be
otherwise directed, we can fix it. However, I think we are *trying* to
send what you want.
--
Bill Cole
bill@scconsult.com or billcole@apache.org
(AKA @grumpybozo@toad.social and many *@billmail.scconsult.com
addresses)
Not Currently Available For Hire
Re[3]: RuleQA low on fuel…
Posted by Tobi <ja...@gmx.ch>.
Bill,
I checked rulesqa and saw that my submission is not there anymore :-( No
idea what my masschecks has as issues. Will check with high Prio after
btk Tomorrow
Cheers
Tobi
------ Originalnachricht ------
Von "Kent Oyer" <ke...@mxguardian.net>
An ruleqa@spamassassin.apache.org; lucilanga@gnome.eu.org
Datum 14.02.2026 01:23:55
Betreff Re[2]: RuleQA low on fuel…
>I'm also a little confused about the masscheck run time. The docs say
>"at or after 9AM UTC". But since this is a daily job, aren't all times
>technically after 9AM? (i.e. 6AM is after 9AM the day before). Perhaps
>it would be better to specify a time range. Job needs to start after X
>but be finished before Y.
>
>
>On Fri, Feb 13, 2026 at 11:24 AM, Lucian Langa <lu...@gnome.eu.org>
>wrote:
>>------------------------------------------------------------------------------------------------------------------
>>CAUTION: External email from: lucilanga@gnome.eu.org
>>Use caution before clicking on links or opening attachments.
>>------------------------------------------------------------------------------------------------------------------
>>Hi Bill,
>>
>>On 13.02.2026 10:57, Bill Cole wrote:
>>>Anyone have ideas for how to fix this?
>>
>>
>>could this once again be that (my) submission time is off (though I
>>might not be the only one)
>>
>>Looking at checkMasscheckContribs.sh time is 15:17 UTC
>>
>>>>SVN tagged rev in nightly_mass_check: 1931846
>>>>
>>>>New masscheck submission listings in the past day:
>>>>SVN rev (Match) File Name (Date)
>>>>1931824 (No) - ham-llanga.r1931824.log (Feb 12 17:26)
>>
>>
>>but rev 1931846 was completed on my side at about Feb 13 18:34 CET
>>(that's 17:34 UTC) so it probably means is was late at least 2 hours.
>>
>>yet on "in the past week" rev 1031743 seems fine at 17:25 (10 mins
>>earlier than 13'th run.
>>
>>>>Recent masscheck submission listings in the past week:
>>>>SVN rev (Match) File Name (Date)
>>>>1931743 (Yes) - ham-net-llanga.r1931743.log (Feb 7 17:25)
>>
>>
>>I must admit never know exactly the time should run my script and to
>>be
>>perfectly honest ".. start your run after H hour" doesn't tell a much.
>>The corpus is variable, the machine running the corpus also a variable
>>and the day of the week (Sat is slower because of the -net rules).
>>
>>It would be much clear (at least for me) if I know the time when I
>>have
>>to rsync my corpus - say make sure your corpus hits the server by H
>>hour. That way I can adapt my starting script time with the corpus
>>size,
>>machine and the day of the week.
>>
>>>See the message below. Also related errors reported to both the
>>>sysadmins and ruleqa lists for over a week.
>>
>>
>>I only got "bad performing rules" mail that doesn't tell much about
>>submission.
>>Can we (ruleqa) get the output of this script too and there
>>was another (I can't remember its name) about the corpus of each
>>contributor and the minimum threshold, that could be also useful.
>>
>>--
>>Cheers,
>>--lucian