Replica in recovering mode taking time to catch up

[jayant]

Hi,


We have a replica member(primary with priority = 20) which went down
due to disk issue and another member with priority 10 takes the role
of primary. After that the down server came up and it becomes
secondary. Initially it recovers very fast but after some it becomes
really really slow.
We are just matching file counts on stale member with primary one and
its recovering slowly.

Our assumption is that once it gets recovered it will become primary.
But the question is as its reaching nearer to primary...recovery
becomes slow. Is this behavior an odd one or correct?

Thanks
Jayant

[Kristina Chodorow]

Is it rolling back ops that weren't replicated before it went down?
You can check for this by grepping for "rollback" in the logs or the
output of rs.status().

Generally the best way to see how well it's catching up is to use:

> db.printReplicationInfo()
> // and
> db.printSlaveReplicationInfo()

in the shell on the primary and secondaries.