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

Re: Slurpd error



Hello,

When I start slapd in master, before slurpd,  with the option -d -1.....

#slapd -d -1

.....
line 51 (replica host=192.168.0.1:389 binddn="cn=Manager,o=univates,c=br" bindmethod=simple credentials=secret)
/usr/local/etc/openldap/slapd.conf: line 51: replica line must appear inside a database definition (ignored)
.....
....
daemon: select: listen=6 active_threads=0 tvp=NULL


But the line of replica is identical in the documentation - http://www.umich.edu/~dirsvcs/ldap/doc/guides/slapd/11.html

When the line of replica is:
replica host=192.168.0.1:389 binddn="cn=Manager,o=univates,c=br" database=ldbm bindmethod=simple credentials=secret


Occur the same line problem.


Dave Horsfall wrote:

On Wed, 12 Mar 2003, Mike Denka wrote:



I discovered the source of the slurpd crash elsewhere and never got back
to the "not mine" mystery, but if anyone else knows the true scoop, I
would also be interested in hearing it.



I've always thought it was because when operating with several replicas, that particular update did not apply to that particular instance.

For example, we replicate to branches all around the world, with each
branch having a slave for everyone else (to cut down on international
traffic).  Because everything gets stuffed into the one replog, it's quite
common to see for example the slurpd thread for Indonesia rejecting an
update meant for USA (it gets picked up by the USA thread).

I eventually got annoyed with this and other problems, and ended up
running multiple slurpds, one per branch.  Yes, we have a weird setup.