Mailing List Archive
tlug.jp Mailing List tlug archive tlug Mailing List Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]Re: [tlug] Email issues
- Date: Sat, 11 Aug 2007 15:09:17 +0900
- From: tlug@example.com
- Subject: Re: [tlug] Email issues
- References: <14178ED3A898524FB036966D696494FB8E4DCC@messenger.cv63.navy.mil> <200708072024.01244.tlug@extellisys.net> <44F1B438-8388-45C4-9740-CF7D1E42B9B8@OnJapan.net>
- User-agent: KMail/1.9.6
On Thursday 09 August 2007 19:30:36 Jim Tittsler wrote: > Which is rather what you would expect given our greylisting implementation. <SNIP> > but just scanning through the greylist I see your mail actually arrives from > more than one IP address, which could impose delays on each. <SNIP> > I think greylisting accounts for the delay. Thank you for the explanation. > I've not seen that big of a delay... I wrote a quick script to analyse the delays based on the timestamps in the email headers (just because I love mathematics/statistics and it seemed like an interesting dataset to look at). I ran it on my TLUG maildir directory, and here are the stats: # samples: 2559 # errors: 298 mean delay time: 15m standard deviation: 56m 17s smallest delay time: 0s median delay time: 37s largest delay time: 22h 3m 37s Note that the errors are emails from Stephen Turnbull, because his SMTP server (uwakimon.sk.tsukuba.ac.jp) has a clock skew. Those emails are ignored for the rest of the stats, and # samples does not include errors. The standard deviation is huge because there are a relatively small number of really large values... Here are the stats with the highest 10% delay times trimmed: # samples: 2304 mean delay time: 4m 18s standard deviation: 7m 49s smallest delay time: 0s median delay time: 27s largest delay time: 29m 30s Here is the data for a histogram (of the trimmed version): 0s ~ 2m 57s 1693 2m 57s ~ 5m 54s 15 5m 54s ~ 8m 51s 256 8m 51s ~ 11m 48s 29 11m 48s ~ 14m 45s 29 14m 45s ~ 17m 42s 32 17m 42s ~ 20m 39s 46 20m 39s ~ 23m 36s 49 23m 36s ~ 26m 33s 70 26m 33s ~ 29m 30s 85 Here is the count for the trimmed samples: 29m 30s ~ 22h 3m 37s 255 The standard deviation is still larger than the mean, but that is to be expected given the nature of the problem. The script used to calculate this data is a quick hack and doesn't even have many comments, but I would be happy to share it if anybody is interested. Cheers, Travis
- Follow-Ups:
- Re: [tlug] Email issues
- From: Josh Glover
- Re: [tlug] Email issues
- From: Stephen J. Turnbull
- References:
- RE: [tlug] Email issues
- From: burlingk
- Re: [tlug] Email issues
- From: tlug
- Re: [tlug] Email issues
- From: Jim Tittsler
Home | Main Index | Thread Index
- Prev by Date: Re: [tlug] [IME] :P Which IME does everyone use?
- Next by Date: Re: [tlug] [IME] :P Which IME does everyone use?
- Previous by thread: Re: [tlug] Email issues
- Next by thread: Re: [tlug] Email issues
- Index(es):
Home Page Mailing List Linux and Japan TLUG Members Links