After I start kmail, continuous network traffic is generated towards one of my imap servers. It's 14k up, 19k down. At first I thought it's for some kind of indexing, but that shouldn't last for several hours. It doesn't stop when kmail is closed, I have to kill the process manually. 6924 ? Sl 11:23 \_ /usr/bin/akonadi_imap_resource --identifier akonadi_imap_resource_1 Reproducible: Always
Created attachment 80068 [details] wireshark-kmail.png I monitored it with Wireshark for a while. It seems that immediately after "client hello" the client closes the connection (FIN), and initiates a new connection. Then comes an "application data" from the server in the old connection for which the client answers with RST (stupid *and* rude). This goes on in an infinite loop until I kill the akonadi_imap process or I have kmail restart it in the settings dialog.
Now I see that I forgot the best part of it: interval mail checking is not enabled, so it shouldn't even do anything in the first place. Especially if kmail is already closed.
I've seen this issue on two different PCs in our office, running Kubuntu (4.10.2) and Gentoo (4.10.4) (both amd64 arch). It appears it started around the time of upgrading to 4.10 for both. It seems to have the same behaviour for both our local mail server (courier, with IDLE enabled) and gmail over imap. Interval checking is enabled on both but looking the longevity of the IMAP connections I'd hazard a guess it's using IDLE on both rather than polling. Peak connection count I've seen in the region of 35/second, which amongst other things caused our mail server, with a normal /var/log/maillog output of 35MB/week to write nearly 1GB in 4 days - each connection was causing one connect and one disconnect line in the log like the following: Jun 6 15:58:57 mail imapd-ssl: Connection, ip=[10.0.11.101] Jun 6 15:58:57 mail imapd-ssl: Disconnected, ip=[10.0.11.101], time=0, starttls=1 As the bug does not seem to present until kmail/kontact has been running for a half an hour to an hour, the most effective workaround I've discovered is to turn the connections offline briefly every 15 minutes with a cron job: export DISPLAY=:0 qdbus org.kde.kontact /KMail org.kde.kmail.kmail.stopNetworkJobs sleep 5 qdbus org.kde.kontact /KMail org.kde.kmail.kmail.resumeNetworkJobs (replacing 'org.kde.kontact' with 'org.kde.kmail2' if you use kmail alone) Occasionally it still goes haywire, to which 'akonadictl restart' appears to be the most effective solution. There is not unusual logging anywhere I can find on the client when the above is happening, even in the console output from the restarted akonadiserver, at least some of which seems to end up on the terminal where you ran the restart command (and I assume is going into ~/.xsession-errors in a clean KDE session).
This seems to me to be related to bug #316840.
(In reply to comment #4) > This seems to me to be related to bug #316840. I'm fairly confident bug 317790 is also describing the same issue, and possibly bug 315382 too.
*** This bug has been marked as a duplicate of bug 316840 ***