Detecting replication delays within a cluster

When a database on one server in a cluster is overdue to be replicated to another server, entries are created in the server log. A group of cluster currency statistics are also available to help proactively spot delays in database replication.

Procedure

Replication delays of four or more hours are reported to the server log. To find these entries, search the log for the string ReplCurrency. Here is an example of a ReplCurrency log entry:
[1240:0004-124C] ReplCurrency Source="discussion.nsf", Dest=CN=Server2/O=Renovations (Q:3) discussion.nsf, 
UpdateToRepl2="07/25/2018 03:12:04 AM", LastRepl="07/24/2018 08:03:28 PM", TimeDiff= 300 min, TimeDiff2=728 min 
Table 1. Information reported in a ReplCurrency log entry
Output field Explanation
ReplCurrency Indicates a replication delay of at least four hours.
Source The database on the current server that is overdue for replication.
Dest The target server.
Q:n SCR queue (out of 4) used to the destination replica for this database. If there are many databases for the same destination queue to the same destination server, a problematic database may be slowing replication for that queue.
UpdateToRep The time that the database was last changed on the destination server.
LastRepl The time that the database was last replicated on the destination server.
TimeDiff The number of minutes since the database was last changed on the current server.
TimeDiff2 The number of minutes since the database on the current server last replicated to the target server. If greater than 5000, the text Warning="5000+ mins on replication is shown.
A repeated ReplCurrency alert for a database can indicate database corruption or a replication problem. In this case, take the following steps to troubleshoot:
  1. Check the replication history to confirm that the warning is valid.
  2. Force a manual replication to see if there are any issues with replication or database corruption.
Note: It's possible to see a ReplCurrency log entry for a database when a new replica is created if the replication currency check runs before the new replica has fully replicated with all the other cluster replicas. Once the initial replication is complete, the log entries stop occurring.
Several statistics are available to report delays in database replication. Enter the following command at the console of a server in the cluster to see the statistics:
show stat replica.cluster.currency.*

A set of currency statistics reports the time frames within which databases last replicated with each other server in the cluster. Statistics are grouped by server.

For example, the following output shows that all 98 replicas the server has in common with the server Mail2/Renovations replicated within one minute:
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Outside8Hours = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Uninitialized = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within15Min = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within1Hour = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within1Min = 98
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within4Hours = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within5Min = 0
[00134C:000008-00000000000004AC]   Replica.Cluster.Currency.Mail2/Renovations.Within8Hours = 0