[Dovecot] v2.0.rc3 released
http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz.sig
A lot of dsync fixes. I think I've fixed now all of the bugs ever reported about dsync and I'm hopeful that it's now stable. All the rest of Dovecot looks pretty good too. Maybe v2.0.0 will be out next week.
Changes since rc2:
* Single-dbox is now called "sdbox" instead of "dbox".
"dbox" will stay as an alias for it for now.
+ Added mail_temp_dir setting, used by deliver and lmtp for creating
temporary mail files. Default is /tmp.
+ doveadm: Added "director map" command to list user -> host mappings.
- imap: Fixed checking if list=children namespace has children.
- director: If all login processes died, director stopped reading
proxy-notify input and caused future login processes to hang
- mail_log plugin configuration was broken
- Using more than 2 plugins could have caused broken behavior
- mdbox: Race condition fixes related to copying and purging
- dsync: Lots of fixesOn Tue, Jul 20, 2010 at 1:57 PM, Timo Sirainen tss@iki.fi wrote:
http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz.sig
A lot of dsync fixes. I think I've fixed now all of the bugs ever reported about dsync and I'm hopeful that it's now stable. All the rest of Dovecot looks pretty good too. Maybe v2.0.0 will be out next week.
Changes since rc2:
* Single-dbox is now called "sdbox" instead of "dbox". "dbox" will stay as an alias for it for now.
+ Added mail_temp_dir setting, used by deliver and lmtp for creating temporary mail files. Default is /tmp. + doveadm: Added "director map" command to list user -> host mappings. - imap: Fixed checking if list=children namespace has children. - director: If all login processes died, director stopped reading proxy-notify input and caused future login processes to hang - mail_log plugin configuration was broken - Using more than 2 plugins could have caused broken behavior - mdbox: Race condition fixes related to copying and purging - dsync: Lots of fixes
Hey do you also want reports on ANY errors/output from dsync? I have a couple mailboxes that I kept getting errors about virtual and/or physical sizes and indexes being corrupt but I just assumed it was a warning or ignorable? If you consider that a bug let me know
On 21.7.2010, at 0.07, Brandon Lamb wrote:
Hey do you also want reports on ANY errors/output from dsync?
I was hoping they'd all be gone in rc3.
I have a couple mailboxes that I kept getting errors about virtual and/or physical sizes and indexes being corrupt but I just assumed it was a warning or ignorable? If you consider that a bug let me know
Were you using NFS? I doubt those are dsync bugs but rather generic index corruption problems and such that dsync just brings to surface because it goes through all mails.
On Tue, Jul 20, 2010 at 4:10 PM, Timo Sirainen tss@iki.fi wrote:
On 21.7.2010, at 0.07, Brandon Lamb wrote:
Hey do you also want reports on ANY errors/output from dsync?
I was hoping they'd all be gone in rc3.
I have a couple mailboxes that I kept getting errors about virtual and/or physical sizes and indexes being corrupt but I just assumed it was a warning or ignorable? If you consider that a bug let me know
Were you using NFS? I doubt those are dsync bugs but rather generic index corruption problems and such that dsync just brings to surface because it goes through all mails.
At one point I was using nfs yes, and also have only been using dovecot lda for 1-2 years so probably lots of old mail with missing/wrong sizes (maildir)
On Tue, Jul 20, 2010 at 11:57 PM, Timo Sirainen tss@iki.fi wrote:
http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz http://dovecot.org/releases/2.0/rc/dovecot-2.0.rc3.tar.gz.sig
A lot of dsync fixes. I think I've fixed now all of the bugs ever reported about dsync and I'm hopeful that it's now stable. All the rest of Dovecot looks pretty good too. Maybe v2.0.0 will be out next week.
I've tried compiling this one on FreeBSD-8. I was running RC2 which had compiled successfully.
mail# gcc -v Using built-in specs. Target: i386-undermydesk-freebsd Configured with: FreeBSD/i386 system compiler Thread model: posix gcc version 4.2.1 20070719 [FreeBSD]
<cut> mv -f .deps/mountpoint.Tpo .deps/mountpoint.Plo /bin/sh ../../libtool --tag=CC --mode=compile gcc -DHAVE_CONFIG_H -I. -I../.. -std=gnu99 -g -O2 -Wall -W -Wmissing-prototypes -Wmissing-d eclarations -Wpointer-arith -Wchar-subscripts -Wformat=2 -Wbad-function-cast -Wstrict-aliasing=2 -MT network.lo -MD -MP -MF .deps/network.Tpo -c -o network.lo network.c libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../.. -std=gnu99 -g -O2 -Wall -W -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wchar- subscripts -Wformat=2 -Wbad-function-cast -Wstrict-aliasing=2 -MT network.lo -MD -MP -MF .deps/network.Tpo -c network.c -fPIC -DPIC -o .libs/net work.o network.c:166: error: duplicate 'static' *** Error code 1
Stop in /usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3/src/lib. *** Error code 1
Stop in /usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3/src/lib. *** Error code 1
Stop in /usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3/src. *** Error code 1
Stop in /usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3. *** Error code 1
Stop in /usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3.
</cut>
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254733744121/+254722743223
"If you have nothing good to say about someone, just shut up!." -- Lucky Dube
On 21.7.2010, at 7.38, Odhiambo Washington wrote:
I've tried compiling this one on FreeBSD-8. I was running RC2 which had compiled successfully.
Yeah, FreeBSD compiling is broken in rc3. Fixed it: http://hg.dovecot.org/dovecot-2.0/rev/4fb7e5327efc
On Wed, Jul 21, 2010 at 2:22 PM, Timo Sirainen tss@iki.fi wrote:
On 21.7.2010, at 7.38, Odhiambo Washington wrote:
I've tried compiling this one on FreeBSD-8. I was running RC2 which had compiled successfully.
Yeah, FreeBSD compiling is broken in rc3. Fixed it: http://hg.dovecot.org/dovecot-2.0/rev/4fb7e5327efc
Thanks. That fixed it. However, out of curiosity, what does the warning below mean?
<snip> s -Wpointer-arith -Wchar-subscripts -Wformat=2 -Wbad-function-cast -Wstrict-aliasing=2 -MT imap-zlib-plugin.lo -MD -MP -MF .deps/imap-zlib-plugin .Tpo -c imap-zlib-plugin.c -o imap-zlib-plugin.o >/dev/null 2>&1 mv -f .deps/imap-zlib-plugin.Tpo .deps/imap-zlib-plugin.Plo /bin/sh ../../../libtool --tag=CC --mode=link gcc -std=gnu99 -g -O2 -Wall -W -Wmissing-prototypes -Wmissing-declarations -Wpointer-arith -Wc har-subscripts -Wformat=2 -Wbad-function-cast -Wstrict-aliasing=2 -module -avoid-version -o lib30_imap_zlib_plugin.la -rpath /opt/dovecot2/lib/ dovecot imap-zlib-plugin.lo ../zlib/lib20_zlib_plugin.la -lrt
*** Warning: Linking the shared library lib30_imap_zlib_plugin.la against the loadable module *** lib20_zlib_plugin.so is not portable! libtool: link: gcc -shared .libs/imap-zlib-plugin.o -Wl,-rpath -Wl,/usr/home/wash/Tools/Dovecot/dovecot-2.0/dovecot-2.0.rc3/src/plugins/zlib/. libs -Wl,-rpath -Wl,/opt/dovecot2/lib/dovecot ../zlib/.libs/lib20_zlib_plugin.so -lrt -Wl,-soname -Wl,lib30_imap_zlib_plugin.so -o .libs/lib30 _imap_zlib_plugin.so libtool: link: ar cru .libs/lib30_imap_zlib_plugin.a imap-zlib-plugin.o libtool: link: ranlib .libs/lib30_imap_zlib_plugin.a libtool: link: ( cd ".libs" && rm -f "lib30_imap_zlib_plugin.la" && ln -s "../lib30_imap_zlib_plugin.la" "lib30_imap_zlib_plugin.la" ) Making all in doc Making all in man </snip>
-- Best regards, Odhiambo WASHINGTON, Nairobi,KE +254733744121/+254722743223
"If you have nothing good to say about someone, just shut up!." -- Lucky Dube
On Wed, 2010-07-21 at 17:08 +0300, Odhiambo Washington wrote:
Thanks. That fixed it. However, out of curiosity, what does the warning below mean?
*** Warning: Linking the shared library lib30_imap_zlib_plugin.la against the loadable module *** lib20_zlib_plugin.so is not portable!
This tries to add imap_zlib plugin to depend on zlib plugin and have it load it automatically. I guess the warning means that it might not work with all libcs. But that's ok, it doesn't have to. If it does work, it just gives a nicer error message if imap_zlib plugin is tried to be loaded without zlib plugin.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Timo Sirainen said the following on 20/07/10 22:57:
Don't really know if it is a bug, probably not.
Even setting "ssl = no" Dovecot tries to open ssl_cert and ssl_key files.
Ciao, luigi
/ +--[Luigi Rosa]-- \
Caution: breathing may be hazardous to your health. -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iEYEARECAAYFAkxHJ00ACgkQ3kWu7Tfl6ZRHNwCgjTkXY9VV9ZA85BlIKkO4ghQo YtkAoItiqtTA3R4dUyv752PRr+jpVHEq =34Kg -----END PGP SIGNATURE-----
On Wed, 2010-07-21 at 18:58 +0200, Luigi Rosa wrote:
Don't really know if it is a bug, probably not.
Even setting "ssl = no" Dovecot tries to open ssl_cert and ssl_key files.
Unfortunately I can't figure out any pretty way to solve this. Adding special casing for ssl_cert/ssl_key/ssl_ca seems evil. This same problem could happen with any setting if "<" is used.
Also handling:
ssl_cert =
would require delaying the "<" processing to see if there's ssl=no.
All in all, seems like too much trouble.
What does the "<" actually do? I just followed the wiki example, and it doesn't clearly explain what it is for.
On 7/21/10 2:39 PM, Timo Sirainen wrote:
On Wed, 2010-07-21 at 18:58 +0200, Luigi Rosa wrote:
Don't really know if it is a bug, probably not.
Even setting "ssl = no" Dovecot tries to open ssl_cert and ssl_key files. Unfortunately I can't figure out any pretty way to solve this. Adding special casing for ssl_cert/ssl_key/ssl_ca seems evil. This same problem could happen with any setting if "<" is used.
Also handling:
ssl_cert =
would require delaying the "<" processing to see if there's ssl=no.
All in all, seems like too much trouble.
-- Steve King
Senior Linux Engineer - Advance Internet, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
Hmm, very interesting. Is this to facilitate functionality such that you may be able to specify the cert and key hashes in the config file with a slightly different syntax?
That is the only thing I could think of using a different syntax like that. I've never seen this kind of syntax in other configs.
On 7/22/10 7:33 AM, Timo Sirainen wrote:
On 22.7.2010, at 6.21, Steven King wrote:
What does the "<" actually do? I just followed the wiki example, and it doesn't clearly explain what it is for. Reads the setting's value from the given file. Similar to how < works in a shell.
-- Steve King
Senior Linux Engineer - Advance Internet, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
It's mainly to support per-ip certificates. Login processes are chrooted and unprivileged, so they can't go opening certificates or their private keys after init. So they need to get the full cert/key from the config process. So config parsing needs to be able to know that it needs to read the given file and put it in a setting instead of just passing the path.
Well, okay, I guess it would have been possible to mark some settings as "read this file" and the "<" wouldn't have been necessary. But the problem with ssl=no would be still the same and it would be less flexible.
On Thu, 2010-07-22 at 13:36 -0400, Steven King wrote:
Hmm, very interesting. Is this to facilitate functionality such that you may be able to specify the cert and key hashes in the config file with a slightly different syntax?
That is the only thing I could think of using a different syntax like that. I've never seen this kind of syntax in other configs.
On 7/22/10 7:33 AM, Timo Sirainen wrote:
On 22.7.2010, at 6.21, Steven King wrote:
What does the "<" actually do? I just followed the wiki example, and it doesn't clearly explain what it is for. Reads the setting's value from the given file. Similar to how < works in a shell.
Ah ok, that makes sense. Thanks for explaining the logic there :o)
On 7/22/10 1:42 PM, Timo Sirainen wrote:
It's mainly to support per-ip certificates. Login processes are chrooted and unprivileged, so they can't go opening certificates or their private keys after init. So they need to get the full cert/key from the config process. So config parsing needs to be able to know that it needs to read the given file and put it in a setting instead of just passing the path.
Well, okay, I guess it would have been possible to mark some settings as "read this file" and the "<" wouldn't have been necessary. But the problem with ssl=no would be still the same and it would be less flexible.
On Thu, 2010-07-22 at 13:36 -0400, Steven King wrote:
Hmm, very interesting. Is this to facilitate functionality such that you may be able to specify the cert and key hashes in the config file with a slightly different syntax?
That is the only thing I could think of using a different syntax like that. I've never seen this kind of syntax in other configs.
On 7/22/10 7:33 AM, Timo Sirainen wrote:
On 22.7.2010, at 6.21, Steven King wrote:
What does the "<" actually do? I just followed the wiki example, and it doesn't clearly explain what it is for. Reads the setting's value from the given file. Similar to how < works in a shell.
-- Steve King
Senior Linux Engineer - Advance Internet, Inc. Cisco Certified Network Associate CompTIA Linux+ Certified Professional CompTIA A+ Certified Professional
participants (5)
-
Brandon Lamb
-
Luigi Rosa
-
Odhiambo Washington
-
Steven King
-
Timo Sirainen