Michael, If you're only concerned with custom metrics, there are a few options. Grok is an htmengine application tailored for use in AWS monitoring and anomaly detection, and because of the AWS requirements may not be suited for you. We also have Taurus, another htmengine application that is specialized for Stock and Twitter data. Taurus is more light-weight and less specialized, implements a basic web api along with the graphite-compatible custom metrics interface, and has no UI. You may want to consider Taurus for your use-case and disable the dynamodb service, and of course, you don't need to run taurus.metric_collectors, since those implement the stock and twitter specific forwarding services, which you don't need.
The other option is to build your own application to fit your specific use-case, starting with htmengine, which is a matter of providing a configuration and implementing an API of your choosing. I believe Matt is working on such a tutorial. On Fri, Jun 12, 2015 at 10:51 AM, Greg Diamos <[email protected]> wrote: > > — > Sent from Mailbox <https://www.dropbox.com/mailbox> > > > On Fri, Jun 12, 2015 at 10:14 AM, Michael Parco < > [email protected]> wrote: > >> I was able to resolve this issue and get grok to launch as it should... >> There were a lot of tiny configs that needed to be altered that I think the >> community would benefit from. Most of them were in the confs and changing >> paths that would mirror the paths when grok is installed on an AWS instance. >> >> I have the grok UI currently up, but I am unable to navigate around >> without AWS credentials (access and secret key) if grok is to be used >> locally is there a way to disable this authentication? If I am looking to >> just use the grok server and collect and push custom metrics to grok >> shouldn't there be a way to auth locallly or not at all? >> >> On Fri, Jun 12, 2015 at 11:12 AM, Austin Marshall <[email protected]> >> wrote: >> >>> >>> On Jun 12, 2015, at 7:02 AM, Michael Parco < >>> [email protected]> wrote: >>> >>> **** WARNING: you are running uWSGI as root !!! (use the --uid flag) **** >>> *your processes number limit is 38338* >>> *your memory page size is 4096 bytes* >>> *detected max file descriptor number: 1024* >>> *lock engine: pthread robust mutexes* >>> *thunder lock: disabled (you can enable it with --thunder-lock)* >>> *Listen queue size is greater than the system max net.core.somaxconn >>> (128).* >>> >>> >>> This is the issue. The default uwsgi listen queue size in grok is >>> hard-coded in >>> https://github.com/numenta/numenta-apps/blob/master/grok/conf/supervisord.tpl#L62 >>> at >>> 1024. >>> >>> You can either bump up the value of net.core.somaxconn on your system to >>> match, or update your supervisor config to go with something lower. >>> >>> On that note, it’s likely others will run into this issue. Making this >>> configurable with a sane default would make a good contribution from the >>> community. >>> >>> >>> grok-supervisord.log >>> 2015-06-12 09:55:34,599 INFO success: metric_listener entered RUNNING >>> state, process has stayed up for > than 1 seconds (startsecs) >>> 2015-06-12 09:55:34,599 INFO success: anomaly_service entered RUNNING >>> state, process has stayed up for > than 1 seconds (startsecs) >>> 2015-06-12 09:55:34,599 INFO success: metric_storer entered RUNNING >>> state, process has stayed up for > than 1 seconds (startsecs) >>> 2015-06-12 09:55:34,599 INFO success: model_scheduler entered RUNNING >>> state, process has stayed up for > than 1 seconds (startsecs) >>> 2015-06-12 09:55:34,632 INFO exited: grok-api_00 (exit status 1; not >>> expected) >>> >>> >>> >> >
