IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
What we now do is have the refresh function and module be on a
seperate object to the schema, only referring to the data and
not excuting on the original ldb and event loop.
That is, we never use another ldb context when calling the
refresh function, by binding the refresh handler to the
ldb and not the schema.
Andrew Bartlett
Change-Id: I5c323dda743cf5858badd01147fda6227599bc16
Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
The issue is that the DN contains a pointer to the ldb it belongs to,
and if this is not kept around long enough, we might reference memory
after it is de-allocated.
Andrew Bartlett
Change-Id: I040a6c37a3164b3309f370e32e598dd56b1a1bbb
Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
Change-Id: I88ee188c776364fd66da388ce01fc9288aa2ded0
Signed-off-by: Andrew Bartlett <abartlet@samba.org>
Reviewed-by: Andreas Schneider <asn@samba.org>
The loadparm context on the schema DB might have gone away already.
Pre-cache the schema refresh interval at load time to avoid worrying
about this.
Andrew Bartlett
This was found based on a log provided by Ricky Nance
<ricky.nance@weaubleau.k12.mo.us>. Thanks Ricky!
In that log, over 2.5 days this particular allocation was repeated:
1715099 talloc_new: ../source4/dsdb/samdb/ldb_modules/schema_load.c:120 contains 0 bytes in 1 blocks
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Fri Aug 17 06:21:18 CEST 2012 on sn-devel-104
This reworks dsdb_replicated_objects_commit() to have a proper local tmp_ctx and
to be more careful about what schema is set (only setting a global schema if
the original schema was global).
In particular, the new working_schema is not given a talloc reference
to the old schema. This ensures that the old schema can go away when
no longer used.
Andrew Bartlett
Autobuild-User(master): Andrew Bartlett <abartlet@samba.org>
Autobuild-Date(master): Sat Aug 11 10:31:57 CEST 2012 on sn-devel-104
This way we do not track both the partition seq number and the
replication USN for schema reload purposes.
We only need one indication of actual data change, and the replication
per-partition sequence number is no more expensive to obtain than the
ldb per-partition sequence number.
Andrew Bartlett
If the value has changed then reload the schema, this means that now the
schema is only reloaded on a periodical basis or if we have been asked
explicitly to do it and not necesserly if the schema partition has
changed.
Samba 4 use to try to reload the schema every time dsdb_get_schema was
called (which could be 20+ time per ldb request). Now we only reload at
most every xx seconds (xx being the value of dsdb:"schema_reload_interval"
or 120). The timestamp of the last reloaded schema is kept in the
dsdb_schema object. There is also a timestamp in the ldb_context, that
is used by the LDAP server to know if it has to reload the schema after
handling the request. This is used to allow that the schema will be
immediately reload after a schemaUpdateNow request has been issued, the
reload can't occur in the handling of the LDAP request itself because
we have a transaction autostarted.
this replaces "return LDB_ERR_OPERATIONS_ERROR" with "return ldb_operr(ldb)"
in places in the dsdb code where we don't already explicitly set an
error string. This should make is much easier to track down dsdb
module bugs that result in an operations error.
The schema refresh operation itself starts requests from the top of the LDB
modules stack (see call "dsdb_schema_set_attributes" - search operations).
This doesn't work well when these do perform "dsdb_get_schema" calls. Since the
new schema isn't marked as "refreshed" atm (but in fact it still is - we didn't
terminate the reload/refresh yet) we could perform other calls to
"dsdb_schema_refresh" and run into serious trouble (segfault).
This commit reworks Samba4's schema loading code to detect when it
needs to reload the schema. This is done by watching the @REPLCHANGED
special DN.
The reload happens by means of a callback, which is only set when the
schema is loaded from the ldb - not when loaded from an LDIF file or
DRS.
We also rework the global schema handling - instead of storing the
pointer to the global schema in each ldb, we store a flag indicating
that the global schema should be returned at run time. This makes it
much easier to switch to a new global schema.
Andrew Bartlett
This choses an appropriate talloc context to attach the schema too,
long enough lived to ensure it does not go away before the operation
compleates.
Andrew Bartlett
This loads the defaultObjectCategory DN as an extended DN, so we can
apply it, with the associated GUID, when setting this on records in
the objectClass module.
Previously we would not store the extended DN components for
objectCategory.
Andrew Bartlett
By splitting the module this way, we can load the schema at startup, after
the partitions module is operational, but we leave the 'mess with details of
entries in the partitions' module to operate only on the partitions module.
Loading the schema later allows us to set the @ATTRIBUTES correctly on all
the databases.
Andrew Bartlett
