[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: mistaken debbugs-submit discard

From: Bob Proulx
Subject: Re: mistaken debbugs-submit discard
Date: Fri, 17 Jan 2014 16:48:56 -0700
User-agent: Mutt/1.5.21 (2010-09-15)

Karl Berry wrote:
>     I give 6 minutes in find
> I know that.  I looked at the bug-coreutils queue in far less than six
> minutes after hitting S, I promise.  More like 30 seconds.  
> FWIW, I think six minutes is a good delay.

Okay good.  I won't do anything.

> Also (though not relevant to the case at hand): I had the impression
> that the six-minute clock only started ticking after exiting caughtspam.
> Is that wrong?  Does it start after hitting S?

Yes.  The 'S' saves the file to spam-new.  Therefore the clock starts
ticking at the time the file is written.

> Merely hitting S does not trigger the debbugs intake somehow, does it?
> That would be totally unexpected for me.    

The 'S' saves to spam-new.  Then the background cron task runs very
minute and invokes find to look for files older than six minutes that
it can process.  If the file is old enough then find finds it and
processes it through formail to extract the discard hash and send it off.
When the message was saved it was also marked for deletion from the
current mailbox.  Exiting the current mailbox causes the messages
marked for deletion to be deleted.  The timer starts from when the
file is saved to spam-new.

> I usually race over to the actual mailman queue :) and approve it before
> it can be discarded.  That has worked fine in all previous cases.

And it should continue to work fine.

> And that's what I did this time ... but the message was not there.

I don't know.

> To me the only explanation that the message was already discarded by
> something else (not me), but there seems to be no explanation as to what
> it was.  I looked in lists:/spam/bug-coreutils but it wasn't there.
> Sigh.

I don't know.  Let me brainstorm a moment.  This new ARM system has
noatime enabled.  The old x86 system didn't have noatime specified.
The find is using -newerct based on ctime.  I think that should still
work okay with noatime.  If not then the find would trigger too early.
I will double check.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]