From: Colin <col...@gm...> - 2014-07-14 10:52:11
|
Hi All, Bad news. Disabling TLS on the mta did not resolve the issue. I'm now seeing timeouts after half an hour instead of 400s after increasing the timeout. Time to figure up tcpdump again! All the best, Colin Waring On 12/07/2014 19:32, Colin wrote: > Hi All, > > Good news. Disabling TLS on the mta has resolved the issue completely. > There isn't any idle time on the connections any more and I've > observed a previously affected server (unable to deliver a message to > us for a couple of days) send through on its first retry attempt. > > I'm not sure what the issue is as I am using the same Exim config as I > have always used. It could be Exim, it could be ASSP but I'm happy > with TLS off as both are on the same box communicating over the > loopback interface. > > All the best, > Coin Waring > > On 12/07/2014 16:38, Colin wrote: >> Further to this, I've finally managed to capture the communication >> between ASSP and Exim with tcpdump. Unfortunately I forgot to disable >> SSL so I can't see the actual SMTP commands but I can see the behaviour. >> >> ASSP Takes a total of 1.9 seconds to send the message to Exim. The >> last packet is an ACK from Exim to ASSP at 1.905104s. >> >> There is then nothing for 400 seconds until at 401.906762 when Exim >> starts talking again and tells ASSP that it has timed out. >> >> I've turned off TLS in Exim and will rerun the capture to see if I >> can get the unencrypted content of the message. I can't seem to get >> wireshark to decrypt the TLS stream even though I've given it a copy >> of the key. >> >> I've also increased the timeout in Exim on the off chance that ASSP >> is for some reason taking a long time to do something before >> finishing off the conversation. >> >> Now I just need to wait for an affected connection to come back in >> again! >> >> On 11/07/2014 20:31, Colin wrote: >>> Hi again, >>> >>> Further on this - the issue doesn't seem to be there when the server >>> first boots. After I while I start seeing these in the logs, not >>> sure how related they are. >>> >>> 2014-07-11 20:14:24 [Worker_10000] Warning: got unexpected signal >>> CONT in Worker_10000: package - main, file - sub >>> main::ThreadMaintMain, line - 457! >>> >>> The problem creeps in and always seems to affect the same senders >>> but the messages do eventually get through. >>> >>> ASSP will not cleanly shut down. If I kill the process and remove >>> the pid file ASSP will start back up, listen on the ports and appear >>> to accept connections but nothing happens or appears in the log. I >>> have to reboot to get the ports back. >>> >>> All the best, >>> Colin Waring. >>> >>> >>> >>> On 10/07/2014 17:13, Nigel Kukard wrote: >>>> On 07/10/2014 09:18 AM, Thomas Eckardt wrote: >>>>>> pipelining >>>>> The pipelining extension has nothing to do with the behavior to >>>>> simply >>>>> deliver multiple emails within one connection. >>>>> >>>>> pipelining is related to SMTP commands not mails - this SMTP >>>>> extension >>>>> makes it possible to send multiple SMTP commands without waiting >>>>> for a >>>>> reply after each one >>>> You're right, I clearly misunderstood the extension :) >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> >>>> Open source business process management suite built on Java and >>>> Eclipse >>>> Turn processes into business applications with Bonita BPM Community >>>> Edition >>>> Quickly connect people, data, and systems into organized workflows >>>> Winner of BOSSIE, CODIE, OW2 and Gartner awards >>>> http://p.sf.net/sfu/Bonitasoft >>>> _______________________________________________ >>>> Assp-test mailing list >>>> Ass...@li... >>>> https://lists.sourceforge.net/lists/listinfo/assp-test >>> >> > |