Great thanks Alexander... We really appreciate your feedback. We (the trace team) are working on solutions to improve this latency. Meanwhile, the eng team would suggest the your use task queue to schedule the cloud trace API call outside of user requests. As Adam suggests...
> >> Second, is there any way to detect which requests AppEngine has sampled >> for tracing so that we only spend the time to report traces for those that >> are sampled for RPC traces? >> > Not sure which runtime the you are is using. In Managed VM, we already make trace context available to external developers as a HTTP header “X-Cloud-Trace-Context” (value format: traceid/spanid;o=options). We are working to expose the same header in Google App Engine. Trace context tells the user whether the request is sampled. Here is the sample code that parses the trace context header. https://github.com/liqianluo/gcloud-trace-java/blob/master/cloud-trace-sdk-java-core/src/main/java/com/google/cloud/trace/sdk/TraceContext.java Hope that helps! ..brad On Sat, Jan 30, 2016 at 4:25 PM, Adam (Cloud Platform Support) < araf...@google.com> wrote: > You could try wrapping the Cloud Trace API call in a push task and > executing it asynchronously at the end of the request which would save you > needing to wait for the URL Fetch to complete inside the request handler. > There isn't a way to 'piggy back' on top of the RPCs used internally (that > is documented at least - you could always try poking around in the ApiProxy > code in the SDK). There exists the RPC API > <https://cloud.google.com/trace/api/> for Cloud Trace however the service > definitions for generating gRPC client code are not yet available. > > On Thursday, January 28, 2016 at 6:14:20 PM UTC-5, Alexander Bertram wrote: >> >> Hi there, >> I've started to look at using the new CloudTrace API to add information >> on CloudSQL requests to traces on AppEngine requests, which we desperately >> need, but I encountered two challenges: >> >> First, even batching all the traces for a request together in a single >> API call adds about 100 ms latency to the request (that's the >> /urlfetch.Fetch bar below), which I don't think I can stomach in >> production. Is there anyway to piggy back custom trace data onto >> AppEngine's existing reporting mechanism, which seems to run independently >> of requests? >> >> Second, is there any way to detect which requests AppEngine has sampled >> for tracing so that we only spend the time to report traces for those that >> are sampled for RPC traces? >> >> Thanks, >> Alex >> >> >> <https://lh3.googleusercontent.com/-NPC894TI-ws/Vqqf5Kna_WI/AAAAAAAABKI/hwK5scVl5mM/s1600/trace.png> >> >> >> -- > You received this message because you are subscribed to the Google Groups > "Google App Engine" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to google-appengine+unsubscr...@googlegroups.com. > To post to this group, send email to google-appengine@googlegroups.com. > Visit this group at https://groups.google.com/group/google-appengine. > To view this discussion on the web visit > https://groups.google.com/d/msgid/google-appengine/0f0433fc-8351-427b-8ae8-8fcdf54d1943%40googlegroups.com > <https://groups.google.com/d/msgid/google-appengine/0f0433fc-8351-427b-8ae8-8fcdf54d1943%40googlegroups.com?utm_medium=email&utm_source=footer> > . > > For more options, visit https://groups.google.com/d/optout. > -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To unsubscribe from this group and stop receiving emails from it, send an email to google-appengine+unsubscr...@googlegroups.com. To post to this group, send email to google-appengine@googlegroups.com. Visit this group at https://groups.google.com/group/google-appengine. To view this discussion on the web visit https://groups.google.com/d/msgid/google-appengine/CAHLXp%3DEKQcs4XppQwtcuJaN8uEbw-bPDb8AfxjWaJb4M_5W91w%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
