[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: SyncRepl - no write access

Turbo Fredriksson wrote:

"Pierangelo" == Pierangelo Masarati <ando@sys-net.it> writes:



   Pierangelo> attrs="* structuralObjectClass entryCSN OpenLDAPACI"

   Turbo> Well, putting that (exactly!) in my syncrepl statement, didn't
   Turbo> replicate ANY (almost) attributes to be replicated.

   Pierangelo> That was just a hint, totally untested :)

Well, looking through the code I see that if I remove the 'attrs' option
all togheter, it does:

               tmp[ n++ ] = ch_strdup( "*" );
               tmp[ n++ ] = ch_strdup( "+" );

So, in theory I should be able to put 'attrs="* +"'. That don't work
(bug!?)... Never mind. For me it works just fine to remove the option
all together (as recomend to me earlier).

Thanx for the help everyone, now everything seems to work. Time will
tell if it's more stable then slurpd :)


I definitely suspect a bug; in fact, I changed test017's slapd.2.conf in HEAD, setting attrs="*,OpenLDAPAci" and the search logged as

conn=12 op=1 SRCH base="dc=example,dc=com" scope=2 deref=0 filter="(objectClass=*)"
conn=12 op=1 SRCH attr=OpenLDAPAci * objectClass structuralObjectClass entryCSN

the two databases are exactly the same (I hadn't any OpenLDAPAci stuff in them, though) with respect to non-operational attribtes; of course, some of the non-generated operational attributes differed, e.g. creatorsName, createTimestamp, modifiersName and modifyTimestamp were missing, and entryUUID was in a different position, with the same value (BTW, I don't understand why the entryUUID is deleted and replaced by and __identical__ value in syncrepl code). If I add the creatorsName, createTimestamp, modifiersName and modifyTimestamp attribtes to the list as well, the two databases are essentially identical. All other operational attributes present in my test, i.e. hasSubordinates and entryDN, are generated.

p.





   SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497