https://issues.apache.org/bugzilla/show_bug.cgi?id=57321

--- Comment #9 from Chaitanya Bhatt <[email protected]> ---
Philippe,

>From a feature design point of view, do you think it would make sense for the
slave instances in distributed mode to directly write "individual sampler
results" to graphite and for cumulated metrics each child instance should
cumulate its results and directly send it graphite as well, leaving Jmeter
master server instance completely free from doing any of these tasks. This
design would be great from a scalability point of view. 

If it is too late to make such changes, then we should at least consider to
make this as an option in the BackendListener, where user could choose to
select Jmeter master to do all the sampler result collection,aggregation and
sending of data to graphite OR could choose an option where all the slave
instances could directly send data to graphite. 

In my opinion, the key to successful adoption of BackendListener is if it
enables Jmeter to overcome its biggest limitation, which is, the ability to
provide a way for the user to collect result metrics without taking a hit on
the performance of Jmeter server itself.

Kindly let me know your opinion. If this is not the right place to carry on
such discussion, please let me know the right forum where I can bring up this
discussion.


Thanks
Chaitanya M Bhatt

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to