questions about getting started with the ClearCase Connector

Hi Ismo,

<Ismo> wrote on 05/15/2008 12:58:33 PM:

some quick points about cc connectors here.
According to instructions, we have to install TC client and
ClearCase client on same machine. I chose jazz server to be that
server, because I do not have separate server right now to offer.
Under <client>\jazz\buildsystem\buildtoolkit\ScmConnectors
is a file syncengine.startup.bat. In it there is line:
set syncuser=ccsync
This I must change in our AD/LDAP environment to some actual domain
username - which must be in clearcase user group as well.

Just to make sure, you do know that this is the Jazz repository user,
not the host OS user used to access ClearCase?
If you have your Jazz repository configured to only allow AD/LDAP users,
then you would have the constraint you mention, but if not, you just
need to identify some random Jazz repository pseudo-user (such as a user
named "ccsync" that you created in the Jazz repository).

So I changed:
set syncuser=<user>.properties -file in this folder:
# Credentials for the Jazz user identity which will perform
ClearCase Connector synchronization
userId=<user> there are three cc elements,
and at first I did not make branch to vobs root \jazz_ucm. But
finally after many errors and thinking, I added it and this finally
solved the problem - I got rest of TC cc connector setup functioning.

It builds a config spec based on the information you gave it (i.e. the
label and the branch), and defaults to /main/0 for elements that have
neither that label or that branch (that is currently stated in the
on-line help ... not sure if that info was in the on-line help in your
version or not). Just for interests sake, for elements that did not
have that label or branch, what version were you expecting it to select?
(We can't use /main/LATEST, because we need to lock the configuration
while we are doing a sync, and it would be very unfriendly to lock /main).

There were error in appearing syncengine cmd window, and it took a
bit thinking again. This was because I used my own EU domain
username and password (in cleartext!) in paakkism.properties -file.
And by this way I got rid of errors in syncengine window. This
showed me that I must be logged on to server with my username and I
also have to use this same username in syncengine.

In production, you would not use your username as the "ccsync" user.
You would create some special account (e.g. "ccsync") that should be
used for synchronization (just like you often need to create a special
account for dbadmin for a database, etc.). To keep from things getting
confused, it would be best if you created a special ccsync account from
the beginning, and you should never login yourself as that account (only
the syncengine should login with that account).

I deleted myself from syncengin jobs and logged on to server as
user-x>. But, because this username had no rights to startup windows
services, such as clearcase ALBD service, I must add <user> to be in
this servers Administrator group. After reboot, I finally go cc sync
engine functioning. Hurraa.

Remember that there are three different user-id's that matter:

- The sync-stream owner: This is a Jazz repository user, who created and
will administer the sync-stream.
- The sync-engine Jazz account: This is a Jazz repository user, which by
default is "ccsync" (but you can pick any id you want). This is the
account the sync-engine uses to talk to Jazz. This should *not* be the
sync-stream owner!
- The ClearCase user: This is a native OS user account (*not* a Jazz
repository user account). This is the account the sync-engine uses to
talk o ClearCase.

The various password files you referred to earlier are all for the
sync-engine Jazz account (by default "ccsync").
For the ClearCase user, that is just the person that fired up the
syncengine.startup.bat script (or in RC3, it is the OS identity of the
person who started the Team Concert client from which they executed the
"restart sync engine" operation.

Now I tried TC connection, and received source codes etc. from
clearcase. This was very enthusiastic moment, but as there has been
issues and after a bit calming, I continued testing. Now I am in
sitation I have to study eclipse (TC) functionality because I
receive error in checkin:
"Change-sets cannot be created on a stream"
I do not know yet what it relates to.

I've never seen that one! Can you describe a bit exactly you were
trying to do when you got that message?
In particular, a screen shot showing the Package Explorer would be
useful, showing the file that you changed.
It sounds like you somehow had a "stream" loaded, instead of a workspace
(you can only create new versions in a workspace).
But I don't know how you would have gotten a "stream" loaded, so that is
just unsubstantiated speculation.
If you can reproduce this bug (or even if not :-), please submit a
defect, describing what you did, and assign it to the SCM category (and
please add me as a subscriber).

Something with streams, though. I guess there is not much to proceed
with this. Pls note in this jyvaskyla_jazz_JAZZ shows this vob
\jazz_ucm and \jazz_ucm\JazzPilotFolder\. ETB -named stuff comes
from separate ETB Subversion!

Yes, Eclipse is happy to let you load projects from many different
source control systems into the same Eclipse workspace.

Cheers,
Geoff