On 13/07/2010 02:41, Andrew Bartlett wrote:
> On Tue, 2010-07-13 at 01:16 +0400, Matthieu Patou wrote:
>> Hello Andrews,
>>
>> As metze and Jelmer are on vacation,
>> can one of you had final look on
>> http://git.samba.org/?p=mat/samba.git;a=shortlog;h=refs/heads/upgradeprovision-review
>> and push it if it's ok.
>>
>> I normally addressed the remarks of Matthias and those made by metze on IRC.
> These look good, but I hope you won't mind if make some comments:
>
> In s4 dsdb: Use the changereplmetadata control
> http://git.samba.org/?p=mat/samba.git;a=commitdiff;h=351ecebb1b294012bc311ff9b38e24a2b6cb77e0
>
> You do a check for
> + objectclass_el = ldb_msg_find_element(res->msgs[0], "objectClass");
> + if (is_urgent&& replmd_check_urgent_objectclass(objectclass_el,
> + situation)) {
> + *is_urgent = true;
> + }
>
> In both arms of the if (rmd_is_provided) statement.
>
> But in any case, if we are offline doing an upgradeprovision, then it
> isn't urgent to replicate anything.
Well even though this control is used only by upgradeprovision right now
it's better to have this just in case ?
This came form the fact that I kept in this branch of the test almost
the same logic.
> In s4 upgradeprovision: do not copy RID Set it's automaticaly created by
> the RID manager
> http://git.samba.org/?p=mat/samba.git;a=commitdiff;h=48773341a19e94ac116bc2aa5327d595c48c4212
> + try:
> + if str(reference[0].get("cn")) == "RID Set":
> + skip = True
>
> Is this the best way to identify the RID Set?
>
obviously objectClass: rIDSet is much better
> Anyway, you have come a very long way on these patches, and I think that
> with these small issues addressed, they look good to go in. The
> upgradeprovision work is really, really important, and I really
> appreciate the effort you keep putting into it.
>
> Thanks,
>
> Andrew Bartlett
>
--
Matthieu Patou
Samba Team http://samba.org