[Dovecot] 0.99.10.7-rc1
Yet another 0.99 release. If it works, tomorrow comes the real one:
+ Added outlook-pop3-no-nuls workaround to fix Outlook hang in
mails with NULs.
+ Config file lines can now contain quoted strings ("value ")
- If client didn't finish downloading a single mail in 30 seconds,
Dovecot closed the connection. This was supposed to work so that
if client hasn't read data at all in 30 seconds, it's disconnected.
- Maildir: LIST now doesn't skip symlinksStrange that I heard about the 30sec bug yesterday for the first time. It's been there forever. Guess people have fast internet connections nowadays :)
Hello,
Timo wrote:
Strange that I heard about the 30sec bug yesterday for the first time. It's been there forever. Guess people have fast internet connections nowadays :)
Actually, have a look at: http://www.dovecot.org/list/dovecot/2004-June/003770.html
I presumed it was the client hanging up w/o dovecot realizing whats going on. Reading that bug description it might have been the other way around, or what does the man who actually can read the code think? ;)
And if so, looking very much forward to 0.99.10.7.deb :)
Regards,
Christian
Christian Balzer Network/Systems Engineer NOC chibi@gol.com Global OnLine Japan/Fusion Network Services http://www.gol.com/
On Tue, 2004-07-13 at 14:51, Christian Balzer wrote:
Actually, have a look at: http://www.dovecot.org/list/dovecot/2004-June/003770.html
I presumed it was the client hanging up w/o dovecot realizing whats going on. Reading that bug description it might have been the other way around, or what does the man who actually can read the code think? ;)
Ah, ok. The abortion was probably because of that, but the error message about locking was because another session was trying to access the same mailbox at the same time.
Hello,
Timo wrote:
On Tue, 2004-07-13 at 14:51, Christian Balzer wrote:
Actually, have a look at: http://www.dovecot.org/list/dovecot/2004-June/003770.html
I presumed it was the client hanging up w/o dovecot realizing whats going on. Reading that bug description it might have been the other way around, or what does the man who actually can read the code think? ;)
Ah, ok. The abortion was probably because of that, but the error message about locking was because another session was trying to access the same mailbox at the same time.
Right. Well, sort of. That lock persists for minutes, certainly longer than 30 seconds. Unless the client realizes that it has been cut off and retries again and again and thus keeps the lock active basically all the time. Alas, if it was dovecot which did cut the connection, shouldn't it have also released the lock at the same time? Or was/is that shortcoming part of the bug?
Regards,
Christian Balzer
Christian Balzer Network/Systems Engineer NOC chibi@gol.com Global OnLine Japan/Fusion Network Services http://www.gol.com/
On 13.7.2004, at 16:07, Christian Balzer wrote:
Ah, ok. The abortion was probably because of that, but the error message about locking was because another session was trying to access the same mailbox at the same time.
Right. Well, sort of. That lock persists for minutes, certainly longer than 30 seconds. Unless the client realizes that it has been cut off and retries again and again and thus keeps the lock active basically all the time. Alas, if it was dovecot which did cut the connection, shouldn't it have also released the lock at the same time? Or was/is that shortcoming part of the bug?
It should have released the lock and the process should have died. So I'm not exactly sure why the error message would come, unless the client actually did try to create many connections at the same time. Luckily this can't happen with 1.0-tests anymore.
participants (2)
-
Christian Balzer
-
Timo Sirainen