confirmed, "login_timeout = 1800" does the trick i still tried "timeout = 1800" last night but did not realize that "login_timeout" could be the trigger
however, sitting behind a dovecot-proxy allows to set here large timeouts to workaround and no more "ERR I'm leaving, you're too slow" seen interesting also for me that debug level 511 generates a 1.9 GB logfile - what is happening on the server in the background that produces so much data for a 20 MB attachment and that only Outlook 2003 seems to trigger this ______________________________ [POP] bindip = 127.0.0.1 port = 110 pop_before_smtp = no banner = POP3 timeout = 1800 login_timeout = 1800 http://dbmail.org/mantis/view.php?id=951 ______________________________ BTW: IMAP: improve mailbox atomicity (71187441a99328cfd585f2dbaccb8618de2a5323) seems also to have closed some memory leaks additional to "plug a couple of memory leaks" (1a5230d1239ae988e11afeb810513373a4eda334) currently imapd consumes around 117 MB after 12 hours running like day and night compared with 1.2 GB after few hours :-) thank you! Am 24.02.2012 09:55, schrieb Paul J Stevens: > This sounds like a known - unconfirmed - bug #951 > > On 02/23/2012 10:20 PM, Reindl Harald wrote: >> Hi Paul >> >> sorry for disturb you again :-( >> but there is still a problem in POP3 >> >> test-attachment created with: >> dd if=/dev/urandom of=attachment.bin bs=1M count=15 >> >> yesterday, before the last POP3 fixes a customer with attachments >> around 8 or 20 MB said he becomes in outlook a error message that >> the server closed the connection shortly before fnisnih - none of >> "our" fixes seems to change this behavior >> _________________________________________ >> >> i can reproduce this still if the connection is slow enough >> and only with Outlook 2003 and large attachments and simulate >> the slow connection with server on VMware Workstation 8.0 >> limit the upstream speed to 1.5 Mbit, clients are even slower >> >> see "receive1.gif" and "receive2.gif" and vmware-settings >> only p...@testserver.rhsoft.net is relevant, other server halted >> interesting is that after the first fail Outlook 2003 starts again >> and also fails after that finally stop trying, sometimes you even >> reveive the whole message, but not commited as OK and still trying >> again >> _________________________________________ >> >> "pop3.c" contains "ERR I'm leaving, you're too slow" >> >> i saw this one time on Outlook 2003 with the 15 MB attachment >> while Outlook 2007 did not have any problem (or had luck), the >> other failed times it gave typical useless microsoft-messages >> saying "the server is not available" >> >> so i guess Outlook 2003 and 2007 doing something different >> >> there must be a bug in handling a client doing nothing >> i monitored the transmission and there was progress and >> data on the line (outlook said 2 minutes left at this speed) >> >> the debug-log would make you not happy because it is >> 1.9 GB large for only this transmission with level 511 >> >> attached "cat dbmail.err | tail -n 10000 > dbmail-log.txt" >> compressed with XZ - most content is binary stuff like this >> and wondering that it does not contain the string "too slow" >> _________________________________________ >> >> Regards >> Harry
signature.asc
Description: OpenPGP digital signature
_______________________________________________ DBmail mailing list DBmail@dbmail.org http://mailman.fastxs.nl/cgi-bin/mailman/listinfo/dbmail
