Bug 717386 - Sending emails usually fails twice before it succeeds (connection to server not being properly closed?)
Summary: Sending emails usually fails twice before it succeeds (connection to server n...
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: evolution
Version: rawhide
Hardware: Unspecified
OS: Unspecified
unspecified
medium
Target Milestone: ---
Assignee: Matthew Barnes
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-06-28 17:53 UTC by Adam Williamson
Modified: 2011-06-29 07:36 UTC (History)
3 users (show)

Fixed In Version: evolution-3.1.3
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2011-06-29 07:36:01 UTC
Type: ---


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
GNOME Bugzilla 653618 0 None None None Never

Description Adam Williamson 2011-06-28 17:53:47 UTC
Since I updated my system to Rawhide (and hence Evo 3.1.2), nearly every time I send an email, I get two failures (and have to hit 'Try Again') before it succeeds:

"The reported error was "MAIL FROM command failed: mail.happyassassin.net Error: timeout exceeded"."

and then:

"The reported error was "MAIL FROM command failed: Connection reset by peer"."

I suspect what's happening is Evo is not correctly closing the connection to the server after it sends a mail, so the next time you try to send a mail it tries to re-use the still-open connection from the last time you sent a mail, and this fails.

The server in question is my own personal server, which is Postfix running on a Mandriva 2011 VM. It may be that I could configure my server somehow to allow the behaviour Evo is trying, but I consider it a bug that Evo regressed from working correctly with my SMTP server configuration in 3.0 to failing like this in 3.1, especially since I can't seem to find any preference to control this behaviour. I can grab logs from the server end if that would help.

(The new 'IMAP+' protocol seems to have similar issues with my IMAP server, courier-imap running on the same box; if I use it instead of the 'IMAP' protocol, I get errors about exceeding the maximum number of simultaneous connections or something like that.)

Comment 1 Matthew Barnes 2011-06-28 18:32:03 UTC
(In reply to comment #0)
> I suspect what's happening is Evo is not correctly closing the connection to
> the server after it sends a mail, so the next time you try to send a mail it
> tries to re-use the still-open connection from the last time you sent a mail,
> and this fails.

Sounds plausible.  If you send a second mail shortly after successfully sending the first one, does the second mail go through without problems?

Comment 2 Adam Williamson 2011-06-29 02:07:15 UTC
"If you send a second mail shortly after successfully sending the first one, does the second mail go through without problems?"

Seems to be the case, yes.

Comment 3 Adam Williamson 2011-06-29 02:11:50 UTC
I'm also seeing stuff like this in my mail server logs:

Jun 28 19:00:53 mailserver postfix/smtp[30298]: E06DE505D4: to=<test
aproject.org>, relay=int-mx.corp.redhat.com[10.4.122.10]:25, delay=0.74, delays=
0.02/0.02/0.49/0.22, dsn=2.0.0, status=sent (250 2.0.0 p5T20r2l032487 Message ac
cepted for delivery)
Jun 28 19:00:53 mailserver postfix/qmgr[6780]: E06DE505D4: removed
Jun 28 19:06:36 mailserver postfix/smtpd[30293]: timeout after END-OF-MESSAGE from unknown[192.168.1.1]
Jun 28 19:06:36 mailserver postfix/smtpd[30293]: disconnect from unknown[192.168.1.1]

which seems to support my conjecture (note the email is sent at 19:00, then the timeout at 19:06).

Other senders - here's my webserver sending out a mail via the same server - seem to disconnect immediately after submitting, even before the message is actually forwarded on:

Jun 28 13:39:12 mailserver postfix/smtpd[29984]: connect from unknown[192.168.1.
4]
Jun 28 13:39:12 mailserver postfix/smtpd[29984]: 1C7FD505E1: client=unknown[192.
168.1.4]
Jun 28 13:39:12 mailserver postfix/cleanup[29988]: 1C7FD505E1: message-id=<20110628203911.161B33AA1F>
Jun 28 13:39:12 mailserver postfix/qmgr[6780]: 1C7FD505E1: from=<apache>, size=1829313, nrcpt=1 (queue active)
Jun 28 13:39:12 mailserver postfix/smtpd[29984]: disconnect from unknown[192.168.1.4]
Jun 28 13:39:19 mailserver postfix/smtp[29989]: 1C7FD505E1: to=<adamwill>, relay=shawmail.vc.shawcable.net[24.71.223.43]:25, delay=7.8, delays=0.41/0.01/0.73/6.6, dsn=2.0.0, status=sent (250 ok:  Message 259242306 accepted)

Comment 4 Adam Williamson 2011-06-29 02:14:05 UTC
While testing the above, I noticed a couple of times where it completely failed to send the message, silently - which is much worse :( It would display one error from the server, I'd hit 'Try Again', and it would go away, and even put the mail in the Sent box - but it wasn't actually sent, as my mail server log confirms. I had to try it again immediately after, and it would work, leaving me with two copies in my Sent box, only one of which really went out...

Comment 5 Milan Crha 2011-06-29 07:36:01 UTC
Thanks for a bug report. I noticed similar behaviour earlier too, but not always. The server probably didn't disconnect so quickly. I moved this upstream, as [1], because this isn't fedora specific. Please see [1] for any further updates, though I've a fix for it, thus this will be fixed in the next release.

[1] https://bugzilla.gnome.org/show_bug.cgi?id=653618


Note You need to log in before you can comment on or make changes to this bug.