Hm, the hardware should be mostly beefier in this new cluster, but there
are a couple of differences: Mainly we are now using RAID instead of JBOD,
and our high volume producers (that 150K / second I mentioned) use TLS.
Also, other producers seem have no problems, it is only MirrorMaker. Even
s
Hi Andrew,
It seems the throughput of the new cluster is smaller than that of the old
cluster. And for this reason MirrorMaker cannot send messages fast enough
(i.e. they expire). I recommend comparing the configurations.
For the hanging MirrorMaker instances, I think looking at stack dumps would
