OpenChange

I'm currently giving a try with samba4.0.0-beta1 in place of alpha 20.

I'll keep you posted.

I can reproduce the issue with latest Samba4 releases.

My preliminary investigations tends to show the issue is related to the way services register to Samba4 new endpoint mapper.
OpenChange is surely missing this at the moment. I'm looking how to fix this.

Following-up on this,

I've updated SOPE, SOGo and that is merely all what was required to pass the initialization bug.

Next, I hit "Cannot open your default e-mail folder" at Outlook 2010 start.

New investigation shows the problem may be related to MAPI handles alignment when pushing mapi response over the wire.

I've identified the origin of the problem:

http://gitweb.samba.org/?p=samba.git;a=commit;h=035342c11719d1daa647c0b2ae7cec27a969f83a

Reverting this patch makes openchange work again properly.

The extensive use of LIBNDR_NOALIGN sometimes combined to LIBNDR_FLAG_REMAINING (made mutually exclusive with the incriminated patch) caused NDR alignment problems for handles.

I'll be look tomorrow at how to fix it in OpenChange.

In the meanwhile OpenChange works perfectly with samba4 beta1 once it is reverted.

May be it is worst to change /script/samba4_ver.sh lpha19 to beta1 or beta2, what do you think ?

And I have investigated 1 more issue in beta2-git but with Openchange installed & run:
1. Samba4 beta2-git and Openchange 4012 are installed;
2. All Samba4 daemons are stopped;
3. I start /usr/local/samba/sbin/samba -d6 -i -M single;
4. I start MS Outlook, it hangs up in stage:
[SOGo: sogo_backend_list_contexts:225]
2012-06-14 03:50:52.374 samba⁷⁸⁶ registered class 'MAPIStoreNotesContext' as handler of 'notes' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreFallbackContext' as handler of 'fallback' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreCalendarContext' as handler of 'calendar' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreOutboxContext' as handler of 'outbox' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreTasksContext' as handler of 'tasks' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreContactsContext' as handler of 'contacts' contexts
2012-06-14 03:50:52.375 samba⁷⁸⁶ registered class 'MAPIStoreMailContext' as handler of 'mail' contexts

then I press Ctrl+C and get

^CJun 14 03:50:59 samba [786]: [ERROR] <0x0x83dc868[LDAPSource]> Could not bind to the LDAP server localhost (389) using the bind DN: cn=Administrator,CN=Users,DC=my,DC=net
Jun 14 03:50:59 samba [786]: <0x094B0378[SOGoMAPIFSFolder]:(null)> directory: /usr/local/samba/private/mapistore/SOGo/test/message/fallback/

then press Ctrl+C again because first pressing wouldn' t provide full exit from run Samba4 / Openchange and get

^CJun 14 03:51:28 samba [786]: [ERROR] <0x0x83dc868[LDAPSource]> Could not bind to the LDAP server localhost (389) using the bind DN: cn=Administrator,CN=Users,DC=my,DC=net

then press Ctrl+C for the third time because previous pressing wouldn' t go into full stop of Samba4 / Openchange and get

Jun 14 03:52:31 samba [786]: [WARN] <0x0x8589518[WOContext]> Got no user from authenticator: <MAPIStoreAuthenticator: 0x87c9108>

after totally about 5-6 attempts to exit from earlier started Samba4 in single and in interactive mode, Samba4 stops, but samba.pid and smbd-filserver.conf.pid remains, is not deleted, both pid files, not only Samba4 daemon pid file.

And it is not important whether I tried to access my shares or not.

If I run /usr/local/samba/sbin/samba -d6 -i -M single and don' t start MS Outlook and it doesn' t following, hang up, I can simply by 1 time pressing of Ctrl+C to stop Samba4 and smbd-filserver.conf.pid is deleted. But anyway samba.pid still remains.

Some log for previous post.

The reason why it hangs here is that you are using a single LDAP server (Samba4's one) with Samba4 running in single process mode.

The LDAP service you're tying to query belongs to the same process currently executing the LDAP request. This is causing the process to hang, indefinitely (or untill timeout) waiting for a reply from a process already busy executing the request.

Oooo1 Oooo1 wrote:

May be it is worst to change /script/samba4_ver.sh lpha19 to beta1 or beta2, what do you think ?

worst or better? ;-)

I'll update to latest release when I fix the problem in trunk.

I thought that even in single mode built-in LDAP server would serve all request as in normal mode (regarding only LDAP requests) .
But if you look at it again the main thing of the issue is nondeletion of pid file, but not samba daemon (samba.pid) but smbd-fileserver.conf.pid file, pid file of smbd daemon.

It was slip of tank:
not worst, but may be better, but in original should be: "May be it is worth to change /script/samba4_ver.sh alpha19 to beta1 or beta2, what do you think ?"

Or "...slip of tunk..." . I hope you have understood so. :)

I have installed Samba4 beta1 + Openchange 4026 + SOPE&SOGo, Dovecot 2.0.19, Exim 4.76, MS Outlook 2010 as client.
At the moment there is serious issue in Samba4+Openchange, as I have found out, regarding with synchronazing or / and failure / not correct of mail folder (or folders) initialization (it is an preliminary investigation) .
It is appeared while MS Outlook are started (works and / or synchronize folders) and looks as the follow:

@Jun 28 00:27:06 samba [5551]: <MAPIStoreMailMessage:0xab04cb0:(null)> -init
Jun 28 00:27:07 samba [5551]: <0x0999C6B0[SOGoMAPIFSMessage]:versions.plist> file '/usr/local/samba/private/mapistore/SOGo/blur/folder/mail/folderSent/versio
ns.plist' new or modified: rereading properties
Jun 28 00:27:07 samba [5551]: [WARN] <MAPIStoreMailMessage:0xab04cb0:24.eml> attempting to get change number by synchronising folder...
Jun 28 00:27:07 samba [5551]: <0x0x9a51408[NGImap4Client]> Note: no key found for sorting, using 'DATE': (null)
Jun 28 00:27:07 samba [5551]: [ERROR] <MAPIStoreMailMessage:0xab04cb0:24.eml> still nothing. We crash!
[2012/06/28 00:27:07, 0] ../lib/util/fault.c:72(fault_report) ===============================================================
[2012/06/28 00:27:07, 0] ../lib/util/fault.c:73(fault_report)
INTERNAL ERROR: Signal 6 in pid 5551 (4.0.0beta1)
Please read the Trouble-Shooting section of the Samba HOWTO
[2012/06/28 00:27:07, 0] ../lib/util/fault.c:75(fault_report) ===============================================================
[2012/06/28 00:27:07, 0] ../lib/util/fault.c:144(smb_panic_default)
PANIC: internal error@

You can find it in two shot logs (new_err_1 & new_err_1a, with some additional info) and in full (big) log (it is in Rar archive, because of big size) .
Also there is other issue, it is appeared sometimes, I have not defined when and what depends on (new_err_2) .
All shot log files are snapshoted from the big log.

Why I think it is first of all Openchange issue, because of after that I can access my Samba4 shared folders and can log in to user mailbox via Web interface and I am still able to send / receive e-mails from Web GUI, in particular from me to me.
Additionally J continued to move forward through this issue and suppose that there are some troubles in synchronization of mail folders located in server at Dovecot and locally placed MS Outlook pst file.
For example deleted some e-mails via Web GUI are still remained in MS Outlook or deleted 2 e-mail from Web GUI results that all e-mails ae de;eted in MS Outlook.

I have understood that the issue is in Sogo code and I will repost this message at Sogo mailing list anyway.
But as I suppose, Sogo doesn' t interract with Samba4 directly, it connects to Samba4 via Openchange.
In this case may be add some code to Openchage - handler exception that could avoid this panic break and such the same or others exceptions that could be in a future.