Suddenly (starting yesterday) my mailserver is very slow when I send mail through it.
I cranked up ettercap on the server to watch what's going on, and it connects immediately when I send mail, but just sits there for between ten and thirty seconds before it finally accepts the mail for sending.
It looks like it's waiting for something to time out. But waiting for what?
The nameserver entries in /etc/hosts are correct.
The only thing that has changed is that I added SPF records to my DNS records a few days ago. They are "v=sfp1 a mx -all". I don't see how that would be related to this situation, though.
The only thing that I can think of that might be relevant is that my IP address (my address, not the server's address) is defined as "theatre" in the server's /etc/hosts file, but a rdns lookup on that address gives this result:
200.59.165.142.in-addr.arpa domain name pointer 142-165-59-200.yktn.static.sasknet.sk.ca.
Could that be the issue? I thought that the system would look as far as /etc/hosts and if it finds an entry there then it will look no further so what the "real" rdns lookup says should be irrelevant.
I tried removing the spam blacklist entries that I have in sendmail.mc (spamcop, spamhaus, etc), thinking that if some of those folks were offline maybe that's what's causing this timeout, but that didn't fix it.
Nothing has changed (other than the SPF entries, but I put them in about a week ago) but my mailserver is slower when I send it mail today than it was yesterday.
Picking up mail, and the webserver stuff are all pretty much instant, just like they have always been.
It may be worth mentioning that sending mail directly from the webserver is still pretty much instant as well. I have a webmail front-end on there (Squirrelmail, actually) and there is no wait when sending mail through it.
What could have happened here?
Somebody in the thread at some point said:
Suddenly (starting yesterday) my mailserver is very slow when I send mail through it.
I cranked up ettercap on the server to watch what's going on, and it connects immediately when I send mail, but just sits there for between ten and thirty seconds before it finally accepts the mail for sending.
On the server, what is in /etc/resolv.conf? What happens if you ssh into the server and try to do, eg
dig <sending client ip>@<nameserver ip in server /etc/resolv.conf>
for example
dig 192.168.0.10@123.123.123.123
and how fast does it come back with a reply?
Are you using a realtime blackhole for lookup? Maybe that is slow.
-Andy
Andy Green wrote:
Somebody in the thread at some point said:
Suddenly (starting yesterday) my mailserver is very slow when I send mail through it.
I cranked up ettercap on the server to watch what's going on, and it connects immediately when I send mail, but just sits there for between ten and thirty seconds before it finally accepts the mail for sending.
On the server, what is in /etc/resolv.conf? What happens if you ssh into the server and try to do, eg
dig <sending client ip>@<nameserver ip in server /etc/resolv.conf>
for example
dig 192.168.0.10@123.123.123.123
and how fast does it come back with a reply?
Are you using a realtime blackhole for lookup? Maybe that is slow.
-Andy
/etc/nsswitch.conf governs the lookup order. The FC6 config was files then DNS, but I've seen other distros that had DNS prioritized over /etc/hosts.
Two other things come to mind: IPv6. There was a discussion recently how to turn off IPv6 in F7 or maybe something is amiss with an upstream nameserver.
I'm not familiar with ethercap, but is anything else going on network-wise while sendmail is sitting there? There is also a debugging mode for sendmail, plus /var/log/maillog might tell you something.
Bob...
On Tue, 28 Aug 2007 13:16:30 -0400 Bob Chiodini rchiodin@bellsouth.net wrote:
I'm not familiar with ethercap, but is anything else going on network-wise while sendmail is sitting there? There is also a debugging mode for sendmail, plus /var/log/maillog might tell you something.
Thanks to both you and Andy for your very helpful replies.
This is a fairly restricted mailserver. While it is reasonably busy, everyone other than me accesses his mail through Squirrelmail -- I am the only one that actually uses a mail client to handle my mail.
While working through your suggestions, checking the configurations and writing an email detailing everything as I found it, it occurred to me that some of the application servers also send mail through the mailserver. (Status reports and the like.) So, I triggered one and watched what happened.
To my surprise, it went through immediately. No pause at all.
That was most interesting, so I sent myself a test email from my main desktop computer (this one) using email (http://email.cleancode.org/) and that went through immediately as well.
That left my mail client as the source of the problem. I downgraded Sylpheed from 2.4.4 to 2.3.1 and the problem went away.
Weird indeed. Especially since I have been using the 2.4.4 version of Sylpheed for the past month with no problems until just the other day.
Sylpheed 2.4.4 still works fine with the sasktel.net mailserver which I believe runs on some sort of Sun gear.
There is still an open question about why it worked for a month and nothing changed except for the fact that it stopped working.
However, at least I know that there is nothing wrong with my mailserver, which was my main concern. And since Sylpheed 2.3.1 still works fine, I guess I'll just stick with that one for the time being.