-
Notifications
You must be signed in to change notification settings - Fork 220
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Slow Jolokia response with cassandra #256
Comments
Which version do you use ? There was an issue with reverse DNS lookups which might be the reason. You can switch this off with a configuration option On the other hand you can use 1.2.3. This 1000 element limit is only a default value (which was set to unlimited in a later version). Setting set the configuration option |
We are using version 1.3.2 at the moment, I also tried 1.3.3 with the same effect. Those are good suggestions, one other thing I was thinking of trying was only specifying the beans that we need. Currently diamond defaults to collecting all of the metrics which potentially there could be a lot of. I am going to try to pare the list down and see if the query times get better. One other anecdote was that I was able to use jmxterm (https://github.com/jiaqi/jmxterm) to query jmx in cassandra directly and it seemed pretty fast to come back, I likely wasnt doing quite as an extensive of a query as jolokia is but perhaps that add some merit to the dns lookup suggestion you made. Thanks so much for the suggestions and I will give them a try and report back if i find anything. regarding the dns issue, the diamond collector i believe is talking directly to 127.0.0.1 so no dns lookup should be required unless jolokia is doing something under the covers. Another interesting annecdote is that jolokia took almost 2 minutes to come back and say my request was invalid for the following:
Also even with maxDepth=1 it still took almost 2 minutes to come back but with less data
|
I am using jolokia 1.3.5 with C* 3.10 and everything works perfectly fast:
|
I am seeing some very slow jolokia request times on my Cassandra 3.0 cluster. Specifically when running the following:
on a slightly loaded cluster I am seeing request times around 30-40 seconds and on a more heavily loaded cluster I am seeing request times in the 2 minute mark. We are currently using jolokia 1.3.2 and v4 of the diamond collector. I also have a Cassandra 1.1 cluster that has the same load and number of nodes and running the same curl command comes back in about 1 second however it is using jolokia 1.2.3 which has a known issue where the keys are truncated at 1000 so my hunch is that is actually what is causing the faster response times.
Is there anything I can do to help diagnose this issue to see what is causing the slowdown or has anyone else experience this?
I have a bug open on the cassandra side as well as i'm not sure which is the culprit here, any advice on how to debug would be greatly appreciated. https://issues.apache.org/jira/browse/CASSANDRA-11677
The text was updated successfully, but these errors were encountered: