Couch Db take 100% CPU usage on replication with heavy data

[biju78]

When we triggering continuous replication with some heavy data( more than 10000 records) we found that couch db process take 100% CPU usage.Most of the time erlang.exe process take more than 60% and many couchjs.exe process take 4% each.
This problem is also occurring without heavy data while we triggering more than 400 continuous replication triggering in a single data base.Once this issue occurs(100% CPU usage) some times replication is too slow or it dose not works.As per some google search we try to resolve this solution by resolving lowering the values of configuration properties(default.ini) worker_processes, worker_batch_size and http_connections. But it dose not works. If any one know how to resolve this issue then please help us..

Thanks in Advance.

[wohali]

There's not enough information to go on here, but this sounds like deployment architecture issues, not a CouchDB bug. Here are some standard recommendations:

• Upgrade to CouchDB 2.1 if you can, replication is improved in this release.
• Try running a 3-node CouchDB 2.1 cluster with a load balancer in front if you are not already.
• Use Mango views and selectors if you can, instead of JavaScript views, for improved performance.

[penkeysuresh]

@wohali I'm facing the same Issue and I posted the logs and error output here #869 . Could you please go through and give some pointers on how to solve this ?

Edit: I've managed to bring back the CPU utilisation to normal levels, it's hardly a solution, but it fixes my problem for time being. Check out the comments here #869