>-----Message d'origine----- >De : [EMAIL PROTECTED] >[mailto:[EMAIL PROTECTED] la part de Rob Arends >Envoyé : lundi 22 mai 2006 14:35 >À : xmail@xmailserver.org >Objet : [xmail] Re: xmail DNS problem : First sample > >I am currently running 1.22 on Win2kSP4 - same as Francis. >I also have in SERVER.TAB >"SmartDNSHost" "10.90.10.150:udp;" > >Previously I had the dreaded 'A' record problem, but now do not. >I WAS running MS DNS, but am now running BIND NAMED in place >of the MS DNS >(on the same box as Xmail). > >There is NO absolute information that proves the move from MS >DNS to NAMED >was the solution, but it has offered me some enhancements, >like 'views', so >I'm unwilling to move back to MS DNS. >
Notice that I sayed that I don't use anymore smartdnshost : In my previous configuration xmail 1.19 + smartdnshost pointing to the local Microsoft DNS I never had this problem. When I updated xmail to 1.21, with smartdnshost intact, I got the problem. After first discussion for this problem on the list, some suggested to not use smartdnshost to avoid use of the local Microsoft dns that could be the problem. So, I removed 'smartdnshost', and now (normally ?) xmail only use its own internal resolver, not the local resolver (yes ? no ?) nor the local microsoft Dns (yes ? no ?) and I have the problem too ... I'm disappointed ... I will try to use 'Bind named', thanks Rob for your tip >I do get the DNS timeouts that Francesco describes. I notice >it mostly when >running DIG & looking for a domain that the server has not >cached. Usually >I re-run DIG immediately, and I get the records. This just >means the my >client gives up before the records are returned to my NAMED >server, and on >to the client - about 2 seconds. I had this same issue with MS DNS. > >This *might* be the root cause of the whole thing. >How much the DNS timeouts affect Xmail I cannot say. > That a good question ;) >Now I'm not saying THAT is EXACTLY what xmail does, but if it >does, that >would go a long way to resolving the "A" record syndrome. >I note that even though I have "SmartDNSHost" set, xmail still >populates >mailroot/dnscache/mx >Possibly confirming my guess work above. >I'd prefer to be able to turn of xmail's dnscache when using >"SmartDNSHost", >as I don't see it enhances performance when "SmartDNSHost" is >used. More >than likely it makes debugging more difficult. > >Perhaps Davide could clarify if Xmail relies on it's cached >dns records in >case of retried sending. > IMHO, it is simply not the 'job' of a mail server program to 'resolve, cache, ...' dns queries ;) and it should let the local os resolver do the job. At least, I don't think caching the resulting dns query for future use really 'enhance' the response time for 'delivery' as in many cases the responding dns server (in most cases local or isp dns) do some cache by itself (especially true when using 'smartdnshost'). >--------------- >On Francesco's comment about the 'old-fashion way'; I agree I liked it >better when you knew there was a problem (your's or >network's), but there >are those 'end-users' that don't care that your link is down >for a short >while - why should that mean they have to send the email again >- can't you >queue it? >And we find ourselves in the situation we are in now - Damned >if you do and >Damned if you don't. > >I have tuned my retries down to about half a day over 10 >retries, rather >than 32 over 2 days. >I find that helps put me in the middle ground between the >"incorrectly typed >address" and the "oops, my link is down" scenarios. > >These are the values I use on xmail cmd line: "-Qt 600 -Qi 2 -Qr 10" > >These are the retries that produces. >>perl zinc.pl 600 2 10 >01 send-time = 0 (00:00:00) next-try = 600 (00:10:00) >02 send-time = 600 (00:10:00) next-try = 900 (00:15:00) >03 send-time = 1500 (00:25:00) next-try = 1350 (00:22:30) >04 send-time = 2850 (00:47:30) next-try = 2025 (00:33:45) >05 send-time = 4875 (01:21:15) next-try = 3037 (00:50:37) >06 send-time = 7912 (02:11:52) next-try = 4556 (01:15:56) >07 send-time = 12468 (03:27:48) next-try = 6834 (01:53:54) >08 send-time = 19303 (05:21:43) next-try = 10251 (02:50:51) >09 send-time = 29554 (08:12:34) next-try = 15377 (04:16:17) >10 send-time = 44932 (12:28:52) next-try = 23066 (06:24:26) Thanks again Rob, I will reduce too the -Q parameters on my xmail server. (but this do not resolve the problem, just reduce the ndr timing ...) > >I hope that helps someone. > >Rob :-) > Yes, it does :) Francis - To unsubscribe from this list: send the line "unsubscribe xmail" in the body of a message to [EMAIL PROTECTED] For general help: send the line "help" in the body of a message to [EMAIL PROTECTED]
