Localisation Error's (German Umlaute)

[tazzy98]

Hi All,

need your help about some localisation problems with Scalix and AD!

1) I manage the users in W2k AD and make a omldapsync on the scalix-machine. Basically this work's fine but if there are any german "umlaute" in the username
or comment the scalixuser is imported/generated with "?" and isn't be usable. The user is also not manageable or deletable in the SAC just over CLI-tools!

2) the absencemessage is also not working with german umlaute

3) The Scalix mailmessages about delivery and/or receipt is in english and should be german

4) The sortingorder in the scalix systemadressbook is wrong users created with last.first and viewed in outlook with first.last although the default sorting in outlook is last.first...

it seems to be a localisationproblem on the scalixmachine but where should i start my search to fix this problems!?

SuSE 10.0 OSS (German) with Scalix 10 SBS
Windows 2000 Server SP4 (German) with Scalix AD Schema Extension


Thanks a lot

[florian]

Tazzy,

thanks for your message. Let me give you some pointers:

1) The first two problems are known issues with SuSE. You probably chose a German locale during installation and it picked one not doing UTF-8, therefore Tomcat is now running in a non-UTF8 environment and gets some character set encodings wrong.

Believe this thread explains best what you need to do: http://www.scalix.com/community/viewtop ... =enus+utf8

2) Server-generated messages depend on the user's Server Language Setting. This can be set with the "-l" option of ommodu or using the user properties in SAC. Obviously, if the non-delivery receipt is coming from the Internet, it's language cannot be controlled and will always be English. Only locally generated Error Messages can be made German.

3) The sort order problem is a bit strange, it should actually sort on the Display Name. Can you provide the output of:

Code: Select all

# grep USRL /var/opt/scalix/sys/general.cfg
# omsearch -s -m CN
# omshowcda -d SYSTEM


However, when it comes to umlaut sort order things, the stuff mentioned in (1) might have an influence as well, so you might want to make sure that your CDA server is also running in a en_US.UTF-8 environment

Cheers,
Florian.

[tazzy98]

thanks florian for your fast answer and suggestions...

On next monday i can test again and i will shortly quote the output!

A few days ago i was in Cologne at the LinuxWorldExpo and i ask my question's at
the Scalix booth and they are all shaking heads with no answer. I guess the guys
i am talking to are all from the sales team
But they had the right tip to ask my questions here in the forum to be fast answered!

By the way florian i think you are german - i am right?
As you can see my english is'nt the best! Is there anywhere a german scalix forum?

Thanks and a nice weekend

[constant]

.
Isn't Florian Dutch?
.

[florian]

I am German and I was in Cologne. And there was Chris, our SE (technical) and at least 3 people working for reseller partners who are techical, hard to believe. Of course, there were a couple of sales people at the booth as well, but we should have had 2 tech guys there at any time.

Whatever.

Hope it works for you.

Cheers from Frankfurt,
Florian.

P.S. There is no German forum. We considered it, however given that we really serve a global market, English is the predominant language for us. German is big, however, it would somewhat split up and dillute forum information and we thought the risks and problems with this outweigh the benefits.

[tazzy98]

Bad Luck we have missed...

It would be nice to talk in real life with "our" community geek

I talked with two guys from pl coteam out of dortmund. The executive technik and sales.
They could not answer my questions and the other guys were busy, possibly you florian i didn't know.
That was in the early morning of the first expo-day.

We talk on the next expo or LinuxTAG...

[tazzy98]

Hi Florian,

here are my results:

I have tested the "UTF8" Problem as you explained in the thread you posted. All the LC Variables was set to POSIX. I fixed this to de_DE.UTF8 for the system itself and also for tomcat by setting LC_ALL to de_DE.UTF8 directly in the setclasspath.sh. The result is negativ. The users with german umlaute are now completely ignored and not imported by the ldapsync.

The outputs you ask for about the sorting problem...

Code: Select all

# grep USRL /var/opt/scalix/sys/general.cfg
USRL_AUTO_GEN_SGI_2_CN=S, G
USRL_AUTO_GEN_AUTHID=S.G@

# omsearch -s -m CN
CN=sxadmin
CN=ScalixUserAdmins
CN=ScalixGroupAdmins
CN=Jupp Schmitz
CN=ScalixUserAttributesAdmins
CN=ScalixAdmins
CN=sxqueryadmin
CN=Klaus Test
CN=sysadm
CN=Hans Trallala
CN=DB-ADM
CN=Software-Entwicklung

# omshowcda -d SYSTEM
Directory name:              SYSTEM
Directory type:              Shared
Sort tag list:               CN/G/I/S
Extract tag list:            @X400-ATTR@/ENTRY-TYPE/DL-MEMBERS/TITLE/ALIAS/EMPL-ORG/EMPL-DEPT/FOREIGN-ADDR/COMMON-NAME/PHONE-1/PHONE-2/HOME-PHONE/HOME-PHONE2/MOBILE-PHONE/ASSISTANT-PHONE/PD-OFFICE-NAME/POSTAL-ADDRESS-STR/CNTRY/L/STATE-OR-PROVINCE/POSTAL-CODE/INTERNET-ADDR/IA-FORMAL/FAX/PAGER-PHONE/ENTRY-DESC
Rebuild Interval (minutes):  15
Last rebuilt at:             20.11.06 16:43:06
Time to Build (minutes):     0
Temporary location:          /var/opt/scalix/tmp
Record block size:           512


I hope it helps...

[florian]

Just to re-verify. Can you run the following command:

Code: Select all

cat /proc/`pgrep java`/environ | tr "\0" "\n" | egrep 'LC|LANG'


On the sort order issue, you obviously set the display name format to Last, First AFTER creating the users. The display name format is only relevant for new users - the current display name can be seen in the CN attribute from the directory and this is what it will sort on, so the behaviour is correct - even though maybe not what you want.

You can probably change the display name manually as these are not so many users.

However, to do it automatically, you would use a mini piece of shell script like this:

Code: Select all

omshowu -m all | while read name; do ommodu -n "$name" -m CN=; done


This should then be followed by an

Code: Select all

omexeccda -d SYSTEM


to regenerate the address book index. Restart Outlook and see if it changes things.


Let me know what you find,
Florian

[tazzy98]

Hi Florian,

sorry for my late answers, but i am in touch with a big SAN-Installation too...

Code: Select all

cat /proc/`pgrep java`/environ | tr "\0" "\n" | egrep 'LC|LANG'

when fresh rebootet the output looks like this:
LC_ALL=de_DE.UTF-8 and just this!!

when i pkill java and then starting tomcat again out of the shell:
LC_ALL=de_DE.UTF-8
LANG=de_DE.UTF-8


is this relevant?

About the sort order issue, we are on the right way!
I am importing/creating the scalix-users out of the ActiveDirectory and there is the sort order per default First,LAST and in the sac i couldn'd edit this for imported users!
So i have to change the sort order in the W2k-AD-Schema. That could be not so easy!!

[florian]

well, i would prefer the first output to include the LANG variable as well, just to be sure. It does on my test system.

In the meantime, could you try to create a user using SAC when the system is in the second state, i.e. after the Tomcat restart and see if that user has correct umlauts?

Florian.

[tazzy98]

User created in SAC with german umlaute works! I created "Jupp Häßler" in SAC and
the email is adapted to "haessler.jupp..." thats perfekt!
But the problem is to import a user from AD with german umlaut now this users would completly ignored to import by omldapsync!?

[florian]

Hm.

It's working in a number of installations, I can assure you of that.

So what remains?
- does AD have any settings to influence character set? IMHO it shouldn't as the LDAP RFCs specify that UTF-8 must be used.
- what are the LANG and LC_ variables in the environment from which you run omldapsync like?

Thx,
Florian.

[tazzy98]

- does AD have any settings to influence character set? IMHO it shouldn't as the LDAP RFCs specify that UTF-8 must be used.

Standard german W2k Installation including SP4 the only changes in the AD are the Scalix schemaextensions!

- what are the LANG and LC_ variables in the environment from which you run omldapsync like?

LC_ALL=de_DE.UTF-8

[florian]

can you set LANG as well before you run omldapsync?

-- f.

[tazzy98]

sorry florian my last info about the LANG variables in the environment from omldapsync was wrong!

in this environment all LC_* variables and LANG set to de_DE.UTF-8 just LC_ALL is empty!
And in the tomcat environment i set LANG and LC_ALL to de_DE.UTF-8!

The correspending users in the AD with german umlaut are still not imported by omldapsync with error (OM 16078 ....invalid character)