Hi,

Here are some of my experiences with MirrorMaker, but I'm also eager to read 
what others do:

1. Main issue for me is rebalancing. If you have several instances of MM under 
the same group, when one of them dies, loses network connectivity, or you just 
need to add new partitions to whitelist/blacklist - the whole thing starts 
rebalancing, which can take quite a bit of time (depending on the number of MM 
instances in a group) during which time the messages are either not being 
mirrored or are sent sporadically. Initially I had one MM group to mirror from 
3 remote clusters into main cluster, which proved to be a big mistake - when 
one cluster would be unreachable, others would suffer. Now I run separate 
groups for each cluster and thinking of drilling it down even further and 
splitting it by topic.

2. Generally MM is pretty straight forward. At least the way I do it, is set 
num.streams to the number of partitions in the remote cluster and num.producers 
to the number of partitions in the destination cluster. The only thing to be 
cautious about is num.consumer.fetchers parameter, because it doesn't do what 
documentation says. See https://issues.apache.org/jira/browse/KAFKA-2008

3. As for monitoring, I track the number of incoming messages into source 
cluster and the same for destination cluster. Then graph them on the same panel 
- as long as the two are about the same, it's a good indicator mirroring is 
working fine.

Karolis



Best Regards,   
[cid:adform-logo-signature_b49a8980-6dcb-4066-8392-780ab0a91ccf.png]  
<http://site.adform.com/>
Karolis Pocius
IT System Engineer
Email: [email protected]<mailto:[email protected]>
Mobile: +370 620 22108
Sporto g. 18, LT-09238 Vilnius, Lithuania
Adform Insider News<http://blog.adform.com/>


[cid:iab_82fe09f0-2828-4e7b-909e-b5e19c3cd91e.png]



Disclaimer: The information contained in this message and attachments is 
intended solely for the attention and use of the named addressee and may be 
confidential. If you are not the intended recipient, you are reminded that the 
information remains the property of the sender. You must not use, disclose, 
distribute, copy, print or rely on this e-mail. If you have received this 
message in error, please contact the sender immediately and irrevocably delete 
this message and any copies.
On Thu, 2016-12-15 at 14:28 +0000, Greenhorn Techie wrote:

Hi,

Good Afternoon.

We are implementing Kafka MirrorMaker to replicate data from Production Kafka 
cluster to DR Kafka cluster. I'm trying to understand answers to the following 
queries:

what are the known bottlenecks / issues one needs to be aware from a 
MirrorMaker perspective.
Also are there any documented or accepted best practices while using 
MirrorMaker?
How to monitor MirrorMaker replication jobs to identify any issues during the 
job execution there-by alerting the DevOps/Support team?

Would be grateful to hear opinions from experts out there.


Thanks

Reply via email to