using zipkin with elasticsearch

[tumux]

Hello,

I've started zipkin on kubernetes with default mysql storage. It works fine. But - I do not rilly like that I have to configure each service to query zipkin (put spans). + mysql - is very slow (started zipkin few days ago, configured 20% of services currently in production and already 500k spans with performance problems). In my environment all the logs from pod's in kubernetes are already beeing forwarded to elasticsearch storage. All those logs are with trace and span info (because there is spring-cloud-sleuth is beeing used).

So I think I already have everything I need in elastic search and this is just a matter of a query to collect info from it and display it by zipkin in analytical view (in forms of traces and spans). Is zipkin capable doing it? Why and do I rilly need separate dedicated storage for zipkin? No matter if it is mysql or something else.

[codefromthecrypt]

This is more an exploration question than a issue on docker. Storage in zipkin serves the api, which primarily serves the zipkin UI. If you choose not to store data in a zipkin storage option, you won't have the option of the UI or dependency graph, or anything derived from it. The MySQL option was noted quite explicitly as not for production, as are all of the images here except for zipkin itself. All of the storage images are test only, in other words. Sorry if somehow you missed this note <https://github.com/openzipkin/zipkin#mysql>, but yeah MySQL is only usable for small setups (like demos) It would seem sensible for you to switch to elasticsearch storage, if you want to take advantage of zipkin, and the efforts folks have put into elasticsearch support. Chat more on https://gitter.im/openzipkin/zipkin if you like and good luck!

…

On Wed, Dec 20, 2017 at 3:43 AM, Tomas Satas ***@***.***> wrote: Hello, I've started zipkin on kubernetes with default mysql storage. It works fine. But - I do not rilly like that I have to configure each service to query zipkin (put spans). + mysql - is very slow (started zipkin few days ago, configured 20% of services currently in production and already 500k spans with performance problems). In my environment all the logs from pod's in kubernetes are already beeing forwarded to elasticsearch storage. All those logs are with trace and span info (because there is spring-cloud-sleuth is beeing used). So I think I already have everything I need in elastic search and this is just a matter of a query to collect info from it and display it by zipkin in analytical view (in forms of traces and spans). Is zipkin capable doing it? Why and do I rilly need separate dedicated storage for zipkin? No matter if it is mysql or something else. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#167>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AAD61956grQyXgrXhBYr68T3g8QmxukNks5tCBHHgaJpZM4RHdiL> .

[tumux]

Thank you for this answer. Looks like my way is to use elasticsearch