syncrepl - when master restarted, slaves loose connection forever (and never reconnects)?
I decided to use syncrepl (refreshAndPersist) instead of slurpd to synchronize.
I made simple tests but something doesn't work right for me.
I started a totally empty OpenLDAP slave - it got replicated from the master. Any change made on the master is also replicated fine.
Unfortunately, when the connection between the slave and the master is broken (i.e., master restarted), the slave never reconnects. According to the fine slapd.conf manual, reconnection should be made quite fast:
"If the connection is lost, the consumer will attempt to reconnect at an interval time (specified by interval parameter; 60 seconds by default) until the session is re-established."
I waited for about 12 hours, and it didn't happen. Restarting slave helped, and all pending changes were transferred from the master.
Am I missing a setting or something?
The slave is running slapd from OpenLDAP 2.3.37, and here is its slapd.conf part concerning syncrepl:
syncrepl rid=0 provider=ldap://master:389 # interval=00:00:00:30 <- also didn't help if it's 30 secs type=refreshAndPersist searchbase="dc=some,dc=domain" bindmethod=simple binddn="cn=replicationuser,dc=some,dc=domain" credentials=secret schemachecking=off
Tomasz Chmielewski wrote:
I waited for about 12 hours, and it didn't happen. Restarting slave helped, and all pending changes were transferred from the master.
Am I missing a setting or something?
The slave is running slapd from OpenLDAP 2.3.37, and here is its slapd.conf part concerning syncrepl:
syncrepl rid=0 provider=ldap://master:389 type=refreshAndPersist searchbase="dc=some,dc=domain" bindmethod=simple binddn="cn=replicationuser,dc=some,dc=domain" credentials=secret schemachecking=off
All right, I read an old manual...
Adding
retry="60 +"
solved the issue...
<quote who="Tomasz Chmielewski">
I decided to use syncrepl (refreshAndPersist) instead of slurpd to synchronize.
I made simple tests but something doesn't work right for me.
I started a totally empty OpenLDAP slave - it got replicated from the master. Any change made on the master is also replicated fine.
Unfortunately, when the connection between the slave and the master is broken (i.e., master restarted), the slave never reconnects. According to the fine slapd.conf manual, reconnection should be made quite fast:
"If the connection is lost, the consumer will attempt to reconnect at an interval time (specified by interval parameter; 60 seconds by default) until the session is re-established."
I waited for about 12 hours, and it didn't happen. Restarting slave helped, and all pending changes were transferred from the master.
Am I missing a setting or something?
Reading slapd.conf(5) correctly ;-)
"In the refreshAndPersist operation, a synchronization search remains persistent in the provider slapd.
Further updates to the master replica will generate searchResultEntry to the consumer slapd as the search responses to the persistent synchronization search.
If an error occurs during replication, the consumer will attempt to reconnect according to the retry parameter which is a list of the <retry interval> and <# of retries> pairs. For example, retry="60 10 300 3" lets the consumer retry every 60 seconds for the first 10 times and then retry every 300 seconds for the next 3 times before stop retrying.
The + in <# of retries> means indefinite number of retries until success."
Gavin.
On Fri, Dec 15, 2006 at 12:15:19PM +0100, Tomasz Chmielewski wrote:
"If the connection is lost, the consumer will attempt to reconnect at an interval time (specified by interval parameter; 60 seconds by default) until the session is re-established."
If an error occurs during replication, the consumer will attempt to reconnect according to the retry parameter which is a list of the <retry interval> and <# of retries> pairs. For example, retry="60 10 300 3" lets the consumer retry every 60 seconds for the first 10 times and then retry every 300 seconds for the next 3 times before stop retrying. The `+' in <# of retries> means indefinite number of retries until success.
syncrepl rid=0 provider=ldap://master:389 # interval=00:00:00:30 <- also didn't help if it's 30 secs
retry="500 +"
type=refreshAndPersist searchbase="dc=some,dc=domain" bindmethod=simple binddn="cn=replicationuser,dc=some,dc=domain" credentials=secret schemachecking=off
WBR
This is ITS#4708, fixed in 2.3.28.
Tomasz Chmielewski wrote:
I decided to use syncrepl (refreshAndPersist) instead of slurpd to synchronize.
I made simple tests but something doesn't work right for me.
I started a totally empty OpenLDAP slave - it got replicated from the master. Any change made on the master is also replicated fine.
Unfortunately, when the connection between the slave and the master is broken (i.e., master restarted), the slave never reconnects. According to the fine slapd.conf manual, reconnection should be made quite fast:
"If the connection is lost, the consumer will attempt to reconnect at an interval time (specified by interval parameter; 60 seconds by default) until the session is re-established."
I waited for about 12 hours, and it didn't happen. Restarting slave helped, and all pending changes were transferred from the master.
Am I missing a setting or something?
The slave is running slapd from OpenLDAP 2.3.37, and here is its slapd.conf part concerning syncrepl:
Howard Chu wrote:
This is ITS#4708, fixed in 2.3.28.
All right. I upgraded to 2.3.30 in the meantime, and added retry="60 +" to slapd.conf.
[syncrepl slave isn't reconnecting] Am I missing a setting or something?
First, I'm concerned that you might be confusing "interval" (used with refreshOnly) with "retry" (used with refreshAndPersist). This has been discussed previously on list, and is described in slapd.conf(5) man page.
With that said, your symptoms sound like ITS #4708, resolved in OpenLDAP 2.3.28. In short, set "retry", upgrade to 2.3.30, and try again.
Aaron Richton wrote:
[syncrepl slave isn't reconnecting] Am I missing a setting or something?
First, I'm concerned that you might be confusing "interval" (used with refreshOnly) with "retry" (used with refreshAndPersist). This has been discussed previously on list, and is described in slapd.conf(5) man page.
I looked at slapd.conf, but on machine running OpenLDAP 2.2.x... That's the confusion...
With that said, your symptoms sound like ITS #4708, resolved in OpenLDAP 2.3.28. In short, set "retry", upgrade to 2.3.30, and try again.
That's what I did, everything works now properly.
PS. does this list lag only for me (a couple of hours), or is it more fundamental? I noticed such lagging a couple of months ago as well.
openldap-software@openldap.org
-
Aaron Richton -
Dmitriy Kirhlarov -
Gavin Henry -
Howard Chu -
Tomasz Chmielewski