Re : slow transport, master.cf and maxproc value
Hello, > I would suggest you grep your mail log for all outbound smtp connections to >these > 10 IP addresses, and document all the recipient domains. Then add all these > domains to your slow transport map. If you still get those 421 errors, >completely > disable concurrency. If you still get the errors, you probably can't fix the >problem > until you upgrade to 2.5+ and have access to rate delay controls. > > Hope this helps. Ok. I'll try to make that. Thank you, -- Myrdhin,
Re: Re : Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
On Mon, Mar 07, 2011 at 11:55:34AM -0600, Stan Hoeppner wrote: > > Unfortunately, the newer postfix rate delay controls are still based on > > the recipient domain, not the MX destination. > > Yeah. He can absolutely fix his current problem with it. But if he has > to configure huge delays to accomplish it, this obviously creates yet > more problems: a backed up queue and mail not delivered in a timely manner. There is a decent chance that once he moves to 2.5 or later, the proposed feedback tuning and larger failure cohort count will solve the issue without rate tuning. The reported issue with the destination is concurrency, not message rate. The OP's mandate to solve the issues with 2.1.5 is rather unfortunate. I'd strongly consider upgrading the legacy systems to 2.7.2 or 2.8.1, after getting a bit of familiarity with the new release on a test machine. -- Viktor.
Re: Re : Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
Noel Jones put forth on 3/7/2011 7:37 AM: > On 3/7/2011 7:21 AM, Stan Hoeppner wrote: >> I would suggest you grep your mail log for all outbound smtp connections >> to these 10 IP addresses, and document all the recipient domains. Then >> add all these domains to your slow transport map. If you still get >> those 421 errors, completely disable concurrency. > > Nice explanation and reasonable advice. Thanks Noel. > >> If you still get the >> errors, you probably can't fix the problem until you upgrade to 2.5+ and >> have access to rate delay controls. > > Unfortunately, the newer postfix rate delay controls are still based on > the recipient domain, not the MX destination. Yeah. He can absolutely fix his current problem with it. But if he has to configure huge delays to accomplish it, this obviously creates yet more problems: a backed up queue and mail not delivered in a timely manner. > Maybe some clever firewall rules could help. I recall Wietse talking about implementing something a while back specifically for this case. IIRC it was something like doing an smtp_bind_address type thing with a bunch smtp clients, each bound to a different virtual IP all on a single NIC, and sending out all the IPs round robin fashion to prevent "bad sender IP reputation". My memory is fuzzy on this. In lieu of this, the best solution I can think of is going multi instance, say 6 extra instances, 7 total. The following is not a complete step by step how-to but an overview of the basic steps required. On the primary instance, remove all current restrictions you've put in place WRT this Orange farm problem. Bind 6 new additional IP addresses as virtual NICs. On the "primary" instance do smtp_bind_address to the current existing IP address, and the same for inet_interfaces. Do this for each of the 6 new instances, so all 7 Postfix instances only use one IP address each. On each of the 6 new instances, leave smtp_connection_cache_on_demand enabled, and set smtp_destination_concurrency_limit to 18. This will allow you to have 18 concurrent smtp connections for each of 10 (primary Postfix instance) active queue destination domains hosted by the Orange farm, 180 maximum smtp connections into the Orange farm. Setup a round robin DNS A record, say "outboundlds.yourdomain.tld" pointing to these 6 new IP addresses. At the edge firewall limit the 6 new IPs to outbound TCP 25 traffic only. Configure each of the 6 new Postfix instances as minimally as possible, and to only accept connections from the IP of the primary instance. On the "primary" instance, you will have a transport_map containing all domains for which you've experienced concurrent delivery problems. orange.fr smtp:outboundlds.yourdomain.tld wanadoo.fr smtp:outboundlds.yourdomain.tld ... smtp:outboundlds.yourdomain.tld ... smtp:outboundlds.yourdomain.tld ... smtp:outboundlds.yourdomain.tld With this setup, you now have essentially an outbound farm of 6 Postfix servers sending from 6 IP addresses to picky receivers. As connection concurrency checking is typically performed by smtpd's on an smtp client IP address basis, and as each of those 10 Orange servers will allow 3 connections per IP address, you will now be able to have 18 concurrent smtp connections to each of the 10 servers in that farm, or 180 total concurrent connections. Currently you can get 30 total connections being they start flogging you. I'd think a peak capacity of 180 concurrent connections should be enough to fix your problem. > Hmm, I'll wonder out loud if a check_recipient_mx_access table that > returns FILTER slow: would help. It would affect all recipients of a > multi-recipient message, but maybe that would be acceptable. I like my idea better. ;) But it is a kludge, and much more complex to implement. I've never done it, so it's possible what I've stated above won't work, and was a big waste of my time combing through postconf5 and Googling. :( It sure would be nice if Postfix would simply round robin outbound smtp connections across all inet_interfaces by default, or have a main.cf option allowing an OP to enable such a thing easily. If such a thing is already in postconf5 I couldn't find it. Using multiple instances to simply accomplish outbound connection balancing across multiple interface IPs is overkill. -- Stan
Re: Re : Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
On 3/7/2011 7:21 AM, Stan Hoeppner wrote: myrdhin bzh put forth on 3/7/2011 1:53 AM: No, it's not SPAM. Ok, agreed, not spam. In fact, zeDomain.tld is a french know domain : wanadoo.fr (and orange.fr). :( The same 10 servers handle mail for both wanadoo.fr and orange.fr. All Postfix controls relevant to your issue available in 2.1.5 are based on connection concurrency, not rate limiting. The connection concurrency is on a *per domain* basis. Orange is the largest ISP in France, yes? How many other recipient domains you send mail to are also hosted by this same Orange 10 server mail farm? If your answer is "a lot", you could completely disable concurrency and still possibly trip their 3 connections per server limit, simply due to the number of recipient domains for which you have mail queued. They didn't list them in the same order, but if you sort them you'll see it's the same 10 servers. smtp.wanadoo.fr.600 IN A 80.12.242.62 smtp.wanadoo.fr.600 IN A 80.12.242.148 smtp.wanadoo.fr.600 IN A 193.252.22.65 smtp.wanadoo.fr.600 IN A 193.252.23.67 smtp.wanadoo.fr.600 IN A 80.12.242.9 smtp.wanadoo.fr.600 IN A 80.12.242.53 smtp.wanadoo.fr.600 IN A 80.12.242.142 smtp.wanadoo.fr.600 IN A 80.12.242.82 smtp.wanadoo.fr.600 IN A 193.252.22.92 smtp.wanadoo.fr.600 IN A 80.12.242.15 smtp-in.orange.fr. 600 IN A 80.12.242.148 smtp-in.orange.fr. 600 IN A 80.12.242.53 smtp-in.orange.fr. 600 IN A 80.12.242.9 smtp-in.orange.fr. 600 IN A 193.252.23.67 smtp-in.orange.fr. 600 IN A 193.252.22.65 smtp-in.orange.fr. 600 IN A 80.12.242.142 smtp-in.orange.fr. 600 IN A 80.12.242.62 smtp-in.orange.fr. 600 IN A 80.12.242.82 smtp-in.orange.fr. 600 IN A 193.252.22.92 smtp-in.orange.fr. 600 IN A 80.12.242.15 I would suggest you grep your mail log for all outbound smtp connections to these 10 IP addresses, and document all the recipient domains. Then add all these domains to your slow transport map. If you still get those 421 errors, completely disable concurrency. Nice explanation and reasonable advice. If you still get the errors, you probably can't fix the problem until you upgrade to 2.5+ and have access to rate delay controls. Unfortunately, the newer postfix rate delay controls are still based on the recipient domain, not the MX destination. Maybe some clever firewall rules could help. Hmm, I'll wonder out loud if a check_recipient_mx_access table that returns FILTER slow: would help. It would affect all recipients of a multi-recipient message, but maybe that would be acceptable. -- Noel Jones
Re: Re : Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
myrdhin bzh put forth on 3/7/2011 1:53 AM: > No, it's not SPAM. Ok, agreed, not spam. > In fact, zeDomain.tld is a french know domain : wanadoo.fr (and orange.fr). > :( The same 10 servers handle mail for both wanadoo.fr and orange.fr. All Postfix controls relevant to your issue available in 2.1.5 are based on connection concurrency, not rate limiting. The connection concurrency is on a *per domain* basis. Orange is the largest ISP in France, yes? How many other recipient domains you send mail to are also hosted by this same Orange 10 server mail farm? If your answer is "a lot", you could completely disable concurrency and still possibly trip their 3 connections per server limit, simply due to the number of recipient domains for which you have mail queued. They didn't list them in the same order, but if you sort them you'll see it's the same 10 servers. smtp.wanadoo.fr.600 IN A 80.12.242.62 smtp.wanadoo.fr.600 IN A 80.12.242.148 smtp.wanadoo.fr.600 IN A 193.252.22.65 smtp.wanadoo.fr.600 IN A 193.252.23.67 smtp.wanadoo.fr.600 IN A 80.12.242.9 smtp.wanadoo.fr.600 IN A 80.12.242.53 smtp.wanadoo.fr.600 IN A 80.12.242.142 smtp.wanadoo.fr.600 IN A 80.12.242.82 smtp.wanadoo.fr.600 IN A 193.252.22.92 smtp.wanadoo.fr.600 IN A 80.12.242.15 smtp-in.orange.fr. 600 IN A 80.12.242.148 smtp-in.orange.fr. 600 IN A 80.12.242.53 smtp-in.orange.fr. 600 IN A 80.12.242.9 smtp-in.orange.fr. 600 IN A 193.252.23.67 smtp-in.orange.fr. 600 IN A 193.252.22.65 smtp-in.orange.fr. 600 IN A 80.12.242.142 smtp-in.orange.fr. 600 IN A 80.12.242.62 smtp-in.orange.fr. 600 IN A 80.12.242.82 smtp-in.orange.fr. 600 IN A 193.252.22.92 smtp-in.orange.fr. 600 IN A 80.12.242.15 I would suggest you grep your mail log for all outbound smtp connections to these 10 IP addresses, and document all the recipient domains. Then add all these domains to your slow transport map. If you still get those 421 errors, completely disable concurrency. If you still get the errors, you probably can't fix the problem until you upgrade to 2.5+ and have access to rate delay controls. Hope this helps. -- Stan
Re : Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
Hello, > From: Stan Hoeppner > > My spammer RADAR is beeping... :) No, it's not SPAM. My client have a domain (for example clientDomain.tld) and a SMTP server. This server rewrite all email addresses from firstname.surn...@clientdomain.tld to firstname.surname-clientdomaine@zedomain.tld (zeDomain.tld is the problem...). My client have some internal lists too. When he send a mail to one of his internal list, for example servi...@clientdomain.tld, this mail is sent to fn1.n-clientdomaine@zedomain.tld, fn2.n-clientdomaine@zedomain.tld, etc. Sometime, the message : smtp.zeDomain.tld[xxx.xxx.xxx.xxx] refused to talk to me: 421 mwinf5c20 ME Trop de connexions, veuillez verifier votre configuration. Too many connections, slow down. OFR004_104 [104] In fact, zeDomain.tld is a french know domain : wanadoo.fr (and orange.fr). :( Wanadoo has changed his configuration last december 2010 and a lot of problem appears (cf in french http://entraide.orange.fr/assistance/messages/index/35320/messagerie-plus-aucun-email-ne-rentre-chez-107-000-utilisateurs-orange-fr.html?dub=2). I'm not the only one who have this problem but orange/wanadoo not responding... -- Myrdhin,
Re: Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
myrdhin bzh put forth on 3/6/2011 5:33 PM: > Really, thank you for your patience :) I have difficulty understanding > perfectly > English... > > > >> No, the *process* limit of "3" is implemented in master(8), which spawns >> processes on demand, up to the process limit. The concurrency limit >> (parallel >> deliveries to a single destination domain) is enforced in the queue-manager. >> The >> queue manager concurrency limit is a maximum, the actual concurrency will be >> lower when the master(8) daemon process limit is reached, or mail arrival is >> not >> high enough to reach peak concurrency. > > So in my conf : > > + transport : > zedomain.tldslow: > > + main.cf : > transport_maps = hash:/etc/postfix/transport > slow_destination_concurrency_limit = 2 > > + master.cf : > slowunix - - n - 3 smtp -o smtp_connection_cache_on_demand=no > > > Can i replace the '3' with a '-' like this : > slowunix - - n - - smtp -o smtp_connection_cache_on_demand=no > > I'm really confused about this configuration... I can't find the right > optimized > configuration... > > To know DNS informations about zedomain.tld SMTP servers, I executed this > command : > > user# dig smtp.zedomain.tld A > > I have 10 responses (round robin) : > ;; ANSWER SECTION: > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x01 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x02 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x03 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x04 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x05 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x06 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x07 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x08 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x09 > smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x10 > > I know (and i tested) each smtp.zedomain.tld servers can only accept 3 max > concurrency connections : 3 for the xxx.xxx.xxx.x01, 3 for xxx.xxx.xxx.x02, > etc. > So I could make up to 30 concurrency connections but i would like to force 3 > per > servers. > > If i configure like this : > + main.cf : slow_destination_concurrency_limit = 30 > + master.cf : slowunix - - n - - smtp -o > smtp_connection_cache_on_demand=no > > is it good ? If i have 22 mails for zedomain.tld, 'smtp's should send mails > to > the smtp.zedomain.tld in this order (with the > smtp_connection_cache_on_demand=no > option): My spammer RADAR is beeping... myrdhin bzh, what is your motivation here? It appears you are trying to setup your single Postfix server to push the maximum amount of mail per second into some number of undisclosed domains that have multiple MX hosts, by getting around their per MX host restrictions. This isn't something white hat mail OPs typically do. If you have a large volume of legit mail for a single domain, and that domain is limiting your delivery rate, causing _problems_ for your systems, you should contact the mail OPs at that domain(s) and _ask_ them how best to address your problem. Attempting to circumvent their countermeasures is a good way to get yourself permanently blacklisted by them. -- Stan
Re : Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
Really, thank you for your patience :) I have difficulty understanding perfectly English... > No, the *process* limit of "3" is implemented in master(8), which spawns >processes on demand, up to the process limit. The concurrency limit (parallel >deliveries to a single destination domain) is enforced in the queue-manager. >The >queue manager concurrency limit is a maximum, the actual concurrency will be >lower when the master(8) daemon process limit is reached, or mail arrival is >not >high enough to reach peak concurrency. So in my conf : + transport : zedomain.tldslow: + main.cf : transport_maps = hash:/etc/postfix/transport slow_destination_concurrency_limit = 2 + master.cf : slowunix - - n - 3 smtp -o smtp_connection_cache_on_demand=no Can i replace the '3' with a '-' like this : slowunix - - n - - smtp -o smtp_connection_cache_on_demand=no I'm really confused about this configuration... I can't find the right optimized configuration... To know DNS informations about zedomain.tld SMTP servers, I executed this command : user# dig smtp.zedomain.tld A I have 10 responses (round robin) : ;; ANSWER SECTION: smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x01 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x02 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x03 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x04 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x05 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x06 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x07 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x08 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x09 smtp.zedomain.tld.471 IN A xxx.xxx.xxx.x10 I know (and i tested) each smtp.zedomain.tld servers can only accept 3 max concurrency connections : 3 for the xxx.xxx.xxx.x01, 3 for xxx.xxx.xxx.x02, etc. So I could make up to 30 concurrency connections but i would like to force 3 per servers. If i configure like this : + main.cf : slow_destination_concurrency_limit = 30 + master.cf : slowunix - - n - - smtp -o smtp_connection_cache_on_demand=no is it good ? If i have 22 mails for zedomain.tld, 'smtp's should send mails to the smtp.zedomain.tld in this order (with the smtp_connection_cache_on_demand=no option): mail #01 -> xxx.xxx.xxx.x01 mail #02 -> xxx.xxx.xxx.x02 mail #03 -> xxx.xxx.xxx.x03 mail #04 -> xxx.xxx.xxx.x04 mail #05 -> xxx.xxx.xxx.x05 mail #06 -> xxx.xxx.xxx.x06 mail #07 -> xxx.xxx.xxx.x07 mail #08 -> xxx.xxx.xxx.x08 mail #09 -> xxx.xxx.xxx.x09 mail #10 -> xxx.xxx.xxx.x10 mail #11 -> xxx.xxx.xxx.x01 mail #12 -> xxx.xxx.xxx.x02 mail #13 -> xxx.xxx.xxx.x03 mail #14 -> xxx.xxx.xxx.x04 mail #15 -> xxx.xxx.xxx.x05 mail #16 -> xxx.xxx.xxx.x06 mail #17 -> xxx.xxx.xxx.x07 mail #18 -> xxx.xxx.xxx.x08 mail #19 -> xxx.xxx.xxx.x09 mail #20 -> xxx.xxx.xxx.x10 mail #21 -> xxx.xxx.xxx.x01 mail #22 -> xxx.xxx.xxx.x02 ??? Thanks -- Myrdhin,
Re: Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
On Sun, Mar 06, 2011 at 10:43:54PM +, myrdhin bzh wrote: > > slow unix - - n - 3 smtp -o > >smtp_connection_cache_on_demand=no -o smtp_destination_concurrency_limit=2 > > >> The second "-o ..." option is pointless and should be removed. > > Ok. > > >> Concurrency limits are enforced in the queue manager, individual delivery > >>agents just deliver one message at a time, it is the queue manager that > >>knows > >>about multiple deliveries in progress. > > So, in my "slow" configuration, the queue manager can only max "create" > 3 'smtp' delivery agents. No, the *process* limit of "3" is implemented in master(8), which spawns processes on demand, up to the process limit. The concurrency limit (parallel deliveries to a single destination domain) is enforced in the queue-manager. The queue manager concurrency limit is a maximum, the actual concurrency will be lower when the master(8) daemon process limit is reached, or mail arrival is not high enough to reach peak concurrency. > But each can only deliver one message at a time and i have only > one domain in my transport table to the 'slow' deliver, so only 2 'smtp' > delivery agents can be created and deliver concurrency. The 3 in my master.cf > slow line is useless ? No, the "-o smtp_destination_concurrency_limit=2" is useless. -- Viktor.
Re : Re : Re : Re : Re : slow transport, master.cf and maxproc value
Thanks Victor, >> You really should be using 2.7.2 or later. I would like to use this version now :) But i can't : it will be possible after servers migrating (waiting 1 month :( ...) > slow unix - - n - 3 smtp -o >smtp_connection_cache_on_demand=no -o smtp_destination_concurrency_limit=2 >> The second "-o ..." option is pointless and should be removed. Ok. >> Concurrency limits are enforced in the queue manager, individual delivery >>agents just deliver one message at a time, it is the queue manager that knows >>about multiple deliveries in progress. Excuse me, I have a little trouble understanding :-s So, in my "slow" configuration, the queue manager can only max "create" 3 'smtp' delivery agents. But each can only deliver one message at a time and i have only one domain in my transport table to the 'slow' deliver, so only 2 'smtp' delivery agents can be created and deliver concurrency. The 3 in my master.cf slow line is useless ? Thank you for your help, -- Myrdhin,
Re: Re : Re : Re : Re : slow transport, master.cf and maxproc value
On Sun, Mar 06, 2011 at 04:21:44PM +, myrdhin bzh wrote: > > From : myrdhin > > Thank you for your help. I'll try your solution. > > Sorry, but i always have "Too many connections, slow down." in my > /var/log/mail.log. > > My Postfix is old (mail_version = 2.1.5) butI am constrained to correct this > problem before doing the migration on another server (with a recent > version of postfix...). The scheduler features that you are new with Postfix 2.5 and required at least Postfix 2.5.7. You really should be using 2.7.2 or later. > Finally i do that : > > # in /etc/postfix/main.cf : > transport_maps = hash:/etc/postfix/transport > slow_destination_concurrency_limit = 2 > > # in /etc/transport/transport : > zeDomain.tld slow: > > # in /etc/postfix/master.cf : > slow unix - - n - 3 smtp -o > smtp_connection_cache_on_demand=no -o smtp_destination_concurrency_limit=2 The second "-o ..." option is pointless and should be removed. Concurrency limits are enforced in the queue manager, individual delivery agents just deliver one message at a time, it is the queue manager that knows about multiple deliveries in progress. -- Viktor.
Re : Re : Re : Re : slow transport, master.cf and maxproc value
Hello :), > From : myrdhin > Thank you for your help. I'll try your solution. Sorry, but i always have "Too many connections, slow down." in my /var/log/mail.log. My Postfix is old (mail_version = 2.1.5) butI am constrained to correct this problem before doing the migration on another server (with a recent version of postfix...). Finally i do that : # in /etc/postfix/main.cf : transport_maps = hash:/etc/postfix/transport slow_destination_concurrency_limit = 2 # in /etc/transport/transport : zeDomain.tld slow: # in /etc/postfix/master.cf : slow unix - - n - 3 smtp -o smtp_connection_cache_on_demand=no -o smtp_destination_concurrency_limit=2 It seems to work : no "Too many connections, slow down." messages in my log file since 2 days but i'm afraid about a mail delivery slowdown... I understand the "smtp_connection_cache_on_demand=no" option but i don't know if the "*_destination_concurrency_limit" option is per process of smtp program. With my master.cf, can i make 6 concurrency connections max on zeDomain.tld ? Or is it only 2 max ? Thanks for your help, -- Myrdhin,
Re : Re : Re : slow transport, master.cf and maxproc value
Thank you for your help. I'll try your solution. -- Myrdhin,
Re: Re : Re : slow transport, master.cf and maxproc value
On Wed, Mar 02, 2011 at 10:58:43PM +, myrdhin bzh wrote: > >> # in master.cf : > > >> slow unix - - n - 5 smtp > > > Something like that. The "5" process limit will mean that with multiple > > "slow" domains at most 5 delivery processes will be used in total for > > all "slow" destinations. This may not be what you want. It can however > > reduce instances of a single MX host getting many connections when it > > serves multiple slow domains. The downside is a potential bottleneck. > > My /etc/postfix/transport file contains one domain (domain.tld). This domain > have 10 SMTP servers (smtp.domain.tld with round-robin DNS configuration). I > know that each SMTP server can only accept 3 concurrency connections. > > I find lot of lines like this in my mail.log : > > Mar 2 20:31:04 localhost postfix/smtp[5394]: 2A5529C4429: host > smtp.domain.tld[xxx.xxx.xxx.xxx] refused to talk to me: 421 mwinf5c20 ME Trop > de > connexions, veuillez verifier votre configuration. Too many connections, slow > down. OFR004_104 [104] > > How can i configure my postfix ? Instead of setting a very low concurrency limit (the default is 20 which should be fine for 10 MX hosts that tolerate 3 connections each), disable demand connection caching for this domain. And also: master.cf: slow unix - - n - - smtp -o smtp_connection_cache_on_demand=no main.cf: slow_initial_destination_concurrency = 3 slow_destination_concurrency_failed_cohort_limit = 3 slow_destination_concurrency_positive_feedback = 1/5 slow_destination_concurrency_negative_feedback = 1/8 This will spread the load more evenly among the 10 (you say equal-weight) MX hosts and will reduce the rate at which concurrency is increased while improving error tolerance so that the destination is not throttled prematurely. -- Viktor.
Re : Re : slow transport, master.cf and maxproc value
>> # in master.cf : >> slow unix - - n - 5 smtp > Something like that. The "5" process limit will mean that with multiple > "slow" domains at most 5 delivery processes will be used in total for > all "slow" destinations. This may not be what you want. It can however > reduce instances of a single MX host getting many connections when it > serves multiple slow domains. The downside is a potential bottleneck. My /etc/postfix/transport file contains one domain (domain.tld). This domain have 10 SMTP servers (smtp.domain.tld with round-robin DNS configuration). I know that each SMTP server can only accept 3 concurrency connections. I find lot of lines like this in my mail.log : Mar 2 20:31:04 localhost postfix/smtp[5394]: 2A5529C4429: host smtp.domain.tld[xxx.xxx.xxx.xxx] refused to talk to me: 421 mwinf5c20 ME Trop de connexions, veuillez verifier votre configuration. Too many connections, slow down. OFR004_104 [104] How can i configure my postfix ? thanks, -- Myrdhin,
Re: Re : slow transport, master.cf and maxproc value
On Wed, Mar 02, 2011 at 10:42:47PM +, myrdhin bzh wrote: > Thanks for your response Victor. > So, if i want max 3 connections (to domain.tld), is this configuration could > work (?) : > > # in main.cf : > transport_maps = hash:/etc/postfix/transport > slow_destination_concurrency_limit=3 Yes. > # in /etc/postfix/transport > domain.tld slow: > > # in master.cf : > slow unix - - n - 5 smtp Something like that. The "5" process limit will mean that with multiple "slow" domains at most 5 delivery processes will be used in total for all "slow" destinations. This may not be what you want. It can however reduce instances of a single MX host getting many connections when it serves multiple slow domains. The downside is a potential bottleneck. -- Viktor.
Re : slow transport, master.cf and maxproc value
Thanks for your response Victor. So, if i want max 3 connections (to domain.tld), is this configuration could work (?) : # in main.cf : transport_maps = hash:/etc/postfix/transport slow_destination_concurrency_limit=3 # in /etc/postfix/transport domain.tld slow: # in master.cf : slow unix - - n - 5 smtp thanks, -- Myrdhin,
Re: slow transport, master.cf and maxproc value
On Wed, Mar 02, 2011 at 11:23:45PM +0100, Myrdhin wrote: > # in main.cf : > transport_maps = hash:/etc/postfix/transport > > # in /etc/postfix/transport > domain.tld slow: > > # in master.cf : > slow unix - - n - 5 smtp -v -o > smtp_destination_concurrency_limit=3 -o slow_destination_rate_delay=1 These settings are in the wrong place. These are qmgr(8) parameters, not smtp(8) delivery agent parameters. Furthermore, rate delay > 0 implies concurrency = 1. > I would like to know if the maximum connections defined in the slow > transport are 3 or 15 (3 smtp_destination_concurrency_limit x 5 maxproc) ? Neither, since you have not exposed the queue manager to these settings. > If it's 15, how could i configure Postfix to make 3 max connections ? If you want a concurrency limit > 1, you can't set a rate limit. All concurrency control is in the queue manager so settings go in main.cf. -- Viktor.