When you do your first write to the index you also append
the Email address to a log file. The mail subroutine also
appends the Email address to a log file. When you run the
script it first checks the length of the files. If the last
open/send failed the index log will be larger. Pop it for
the address missed.
Your mail sub would be something like this:
open sendmail || &retry_then_exit # Let the retry sleep(10)
open mail_log
write address
exit;
Ron
At 08:30 AM 5/12/98 -0700, Jack Killpatrick wrote:
>Rich Kulawiec wrote:
>
>> On Tue, May 12, 1998 at 01:07:21AM +0000, Peter J. Schoenster wrote:
>> > On 11 May 98 at 19:49, Jack Killpatrick wrote:
>> >
>> > > I would like to have a way to check the emails that have been sent
>> > > via sendmail against the ones in the database so I can spot any
>> > > errors that may have occurred in the sendmail process. For example,
>> > > let's say sendmail fails for some reason (some loose and crazy
>> > > process that stymies the system so that sendmail is rendered
>> > > useless, while perl and msql continue to function). I'd like to
>> > > compare a record of sent mail to the records in the database, using
>> > > the index as a key. Then I could spot any submissions that were not
>> > > sent out by sendmail.
>> >
>> > Umm.. not sure if I really understand but that hasn't stopped me
>> > before.
>> >
>> > When you open sendmail do you not do this:
>> >
>> > open sendmail || do something else
>>
>> You're correct, and you're not correct.
>>
>> What you propose will tell you if sendmail returns a non-zero exit
>> code immediately, but that isn't what you really need to know: you
>> need to know if the mail actually got delivered. And there's no
>> way to know that at the time you submit mail to sendmail's queue,
>> because sendmail itself doesn't know if the mail's deliverable (except
>> in truly pathological cases, e.g. malformed addresses).
>>
>> In other words, just because sendmail accepts a message for delivery
>> on your local host doesn't mean that it will be successful. There can
>> be transient failures (DNS timeouts, loss of connectivity to the
>> remote host), permanent failures (unknown host, unknown user) and about
>> a zillion other things that can go wrong.
>>
>> My solution would be to use the -f flag to set the return adddress, e.g.
>>
>> "|/usr/lib/sendmail -oi [EMAIL PROTECTED] someuser@somewhere"
>>
>> as well as burying the true recipient in an X-header, e.g.
>>
>> X-Really-Really-To: someuser@somewhere
>>
>> and then have mail coming back to "bounces" piped into a perl script
>> that extracts the X-Really-Really-To field and marks it as "not
>> delivered" in your database or wherever it needs to be noted.
>
>Yes, that would work for checking the mail _if_ it was in fact sent out.
>What I really need to do is find out if sendmail even sent the mail out. I
>need to know what email send was being attempted when sendmail did not
>respond.
>
>This is a little more on my scenario: I am using a virtual server at
>iServer. They limit the amount of processes that you can run at one time.
>Last week, one day, I was writing some crappy code and maxed out my process
>limit. So, when my scripts tried to use sendmail it didn't work. My client
>called and said, hey, we haven't had any emails today, is there something
>wrong with the system. I found a zillion processes, killed them and then
>sendmail started working again. They then asked me what emails they had not
>received. I looked in my database and gave them copies of the last records
>in the table, but told them that I didn't know when sendmail stopped being
>usable, so they'd have to check the database info against the emails they
>already received to see what they hadn't received. It was kind of a hassle.
>In the first place, I wasn't aware of the problem, in the second place I
>didn't have an easy way of tracking the unsent emails. Note that these
>weren't even emails drifting around in cyberspace; they never even got
>created.
>
>So, I'm looking for a way to be able to know whether an email was created
>and sent or not. Let's say sendmail doesn't even start up, and thus can't
>return an exit code, I need to know what email send (ideally, identified by
>the 'index' variable in my table) was being attempted at that time.
>
>Could my script just look for _no_ exit code coming from sendmail, write an
>error to the user in html and write a record to my database detailing the
>non-send? Seems like that could work. Hmmm. At that point the user on the
>web would get an error notice, but not me (I don't plan on sitting at the
>console). I could ask the web user to call Customer Service at my client and
>notify them of the error, then Customer Service could call me. Hmmm, that
>might work. Any thoughts?
>
>TIA,
>Jack
>
>____________________________________________________________________
>--------------------------------------------------------------------
> Join The Web Consultants Association : Register on our web site Now
>Web Consultants Web Site : http://just4u.com/webconsultants
>If you lose the instructions All subscription/unsubscribing can be done
>directly from our website for all our lists.
>---------------------------------------------------------------------
>
____________________________________________________________________
--------------------------------------------------------------------
Join The Web Consultants Association : Register on our web site Now
Web Consultants Web Site : http://just4u.com/webconsultants
If you lose the instructions All subscription/unsubscribing can be done
directly from our website for all our lists.
---------------------------------------------------------------------
