cYuSeDfZfb cYuSeDfZfb wrote:
> Hi,
>
> We noticed that, when using "slapd-watcher -i X" option to refresh display
> every X seconds, lagging replication statuses are often not cleared, when in
> fact
> replication as already recovered.
>
> In our MMR environment, we often see this for longer periods of time in
> slapd-watcher output :
>
> contextCSN: 20230913104435.605937Z#000000#0dd#000000 actv@2023-09-13
> 10:44:35, idle@2023-09-13 10:44:37
>
> But when running like this:
> do
> timeout --foreground 1 $SLAPDWATCHER -b $BASE -D $LDAPBINDDN -w $ADMINPW
> "${SERVERURIS[@]}" -s ${SERVERIDS[*]}
> done
>
> the lagging replication lines change back to "idle, sync'd" immediately after
> replication has recovered.
>
> It feels like perhaps there is something wrong in the way the -i X option is
> implemented.
There are no bugs in the slapd-watcher tool. But you must list the server URIs
in ascending order of serverID, and
make sure you provide the list of serverIDs in the matching order. If the list
of SIDs is out of order it won't be
able to compare the CSNs to the correct master.
--
-- Howard Chu
CTO, Symas Corp. http://www.symas.com
Director, Highland Sun http://highlandsun.com/hyc/
Chief Architect, OpenLDAP http://www.openldap.org/project/
