Обсуждение: [GENERAL] Lag in asynchronous replication
hi,
in case of automated failover i want to check if slave ia lagging from master and only if it is in sync, i want to do failover. But I am working in a virtual cloud environment so by that time the master VM may not be available to me. How can i check the lag in that case ?
Regards,
SUBHANKAR CHATTOPADHYAY
On Thu, Mar 23, 2017 at 11:37 PM, Subhankar Chattopadhyay <subho.atg@gmail.com> wrote: > in case of automated failover i want to check if slave is lagging from > master and only if it is in sync, i want to do failover. But I am working in > a virtual cloud environment so by that time the master VM may not be > available to me. How can i check the lag in that case ? Is your environment switching dynamically to async if the lag is too important? If not, once you have reached a sync state, the master would wait for all transactions commits to complete on the slave, so once the client has received a commit confirmation you have the guarantee that the data is already flushed on the slave. In this case you don't need to know what happens on the master. -- Michael
Hi Michael,
Are you asking to have slave with synchronous replication?
Regards,
SUBHANKAR CHATTOPADHYAY
On 24 Mar 2017 09:33, "Michael Paquier" <michael.paquier@gmail.com> wrote:
On Thu, Mar 23, 2017 at 11:37 PM, Subhankar Chattopadhyay
<subho.atg@gmail.com> wrote:
> in case of automated failover i want to check if slave is lagging from
> master and only if it is in sync, i want to do failover. But I am working in
> a virtual cloud environment so by that time the master VM may not be
> available to me. How can i check the lag in that case ?
Is your environment switching dynamically to async if the lag is too
important? If not, once you have reached a sync state, the master
would wait for all transactions commits to complete on the slave, so
once the client has received a commit confirmation you have the
guarantee that the data is already flushed on the slave. In this case
you don't need to know what happens on the master.
--
Michael
On Fri, Mar 24, 2017 at 3:11 PM, Subhankar Chattopadhyay <subho.atg@gmail.com> wrote: > Are you asking to have slave with synchronous replication? (top-posting is annoying) No, slaves cannot do synchronous replication. I am just telling that once you are sure that a sync state has been achieved on the master, you have the guarantee that data gets synchronously replicated on the standbys as long as you do *not* change synchronous_standby_names. So there is no actual need to know what's the state of the master during a failover to a sync standby. -- Michael