[Top] [All Lists]

Re: 2821bis consideration - New 2nd attempt Retry Strategy recommendation

2007-11-17 05:30:54

SM wrote:

A sender not familiar with Greylisting may find the rejects strange at first and might blame the software.

We get blame for everything. <g>

I use Greylisting in the DATA phase. Some people prefer it to be done at RCPT TO to reduce processing overhead.

I have to see if our GL people are comfortable enough with GL to justify adding a RCPT TO rejection option.

Capturing content is not workable for us due to the additional disk space required.

Thats probably the only common comment I see now. Deleting this stuff more frequently. My answer? "erase wcgl\msg*.dat" :-)

I have encountered problems with Greylisting. I came across odd cases.

Care to share a few?

Note that the following is merely a comment. You are better placed than me to decide what's best for your environment.

Well, its the default and it was pretty much a SWAG <g> I'm sure operators have fine tuned them, and other SMTP related options, if they had to, but we haven't heard of any issues or concerns with mail delivery so thats most important. And if there were an issue, it would addressed pronto.

We do have a very aggressive system, and thats what they want. Wouldn't have any other way. We all don't have to be the same, right?

You are doing more than three retries over an hour period. That's quite high in my opinion.

Its only high if its problem across the board. I'm not going to worry about the few (Pareto principle). Also, 4 vs 2 is not much a difference, but what does high translate too? Overhead? For whom? Are we dealing with a down system? a session level 45x reject?

I'm of the opinion, that if it reaches the session, the odds are very hight it will succeed in 1 try or 2 tries for most GL systems, maybe the 3rd try with you. :-) Those with higher block times, well, are just playing games with spammers and could less about business customers.

Also, from what I have seen, top ISPs are also well below 30 mins 2nd retry. In fact for, it has a 5 minute 2nd retry. But AT&T took them over, so lets see what servers are doing. ..... Yup, exactly 5 mins 2nd retry which is what I saw (<30 mins) by the majority when I was looking at the time.

IMO, 30 minutes is really too high and IMO, it corresponds to the good old days of less reliable machines, more down systems, less reliable connections, slower bandwidths, less scaled systems, etc, etc. But not today. With high performance machines, multi-threading software, Virtual Machines, higher bandwidths, better scaling, more challenge/response systems, etc, may suggest the x821 30 minute retry "guideline" is outdated. :-)


Hector Santos, CTO

<Prev in Thread] Current Thread [Next in Thread>