From: "Phil Barnett" <[EMAIL PROTECTED]>
On Thursday 28 June 2007 15:22, Lindsay Haisley wrote:
Attached is a proposed patch for /var/lib/spamassassin/rules_du_jour
which addresses the problem of the refresh URL which Rules Emporium
sometimes sends out instead of a valid cf file. Basically, this patch
greps the downloaded file for the string "META HTTP-EQUIV", which should
never occur in a valid rules file, but is part if the refresh URL. If
the downloaded file is a refresh URL, it's deleted, the script waits 1
second and tries again, up to 3 times. If the download fails after 3
tries, the bad file is deleted and the script moves on.
You might try running rules_du_jour from a cron job with the -D option
and redirecting the output to a /tmp file and see if you get any notices
about "Download of .... FAILED after 3 tries", in which case I've
mis-diagnosed the problem somewhat. In any event, the problem file
should be deleted rather than causing a --lint failure in spamassassin.
I'm going to try this, but with a 5 minute wait. I run it in the middle of
the
night anyway, who cares how long it takes.
Actually, the proper response might be a random wait.
A constant one second wait at the end of each fetch loop works just peachy
with a wget fetch. I cannot honestly speak for curl mode. My script does not
use it.
You will have to wait for up to a day for the Prolexic block to go away.
An hour or two worked for me. A friend left it overnight and tried the delay
trick after he got back from work. It's worked for me and for him. Although
he did put in the random length wait. (I have "quite a number" of rule sets
active here - roughty 40.)
{^_^} Joanne, who has been suggesting this for weeks now.
