On 04/03/2012 02:12 AM, Roberto Polli wrote:
Hi Rich,
Rich Megginson<[email protected]>
On RH documentation I read to:
- upgrade all masters;
- then upgrade slaves;
- lately restart.
Yes. This is the recommended procedure.
This approach seems to lead to some service discontinuity, as - during
this migration - I should stop writes to all master/slaves.
Why?
To avoid discontinuity, I should do a rolling restart, right?
Rolling restart means I have a small time frame T0 where schemas are different
between nodes.
If during T0 a write is replicated between NodeB and NodeA, I suppose
replication will fail due to mismatching schema, right?
Yes.
Alternately, you could update the masters first. If you update the
schema over LDAP to your "primary" master, the schema will be replicated
first, before any data that uses the schema. Note that schema
replication is "single" master. That's why you should pick one master
to be your "primary" master for schema updates.
I found a possible typo here:
http://directory.fedoraproject.org/wiki/Dynamically_Reload_Schema
Yes. I have fixed the wiki ... attributetypes, objectclasses,
matchingrules, are now operational
is there a way to tell 389 to print all operational attributes?
In an ldapsearch request? No.
Thx + Peace,
R:
--
389 users mailing list
[email protected]
https://admin.fedoraproject.org/mailman/listinfo/389-users
