I could be wrong, but I don't think Accumulo has changed anything about the way it is emitting metrics in 1.10 that would be substantially different from 1.8. It's possible that the GraphiteSink has changed how it works in your version of Hadoop. Or maybe newer versions of InfluxDB stores things differently or interacts with the Grafana dashboard differently.
It's very hard to say, since none of those projects are owned, developed, or directly supported by the Accumulo PMC. We just emit metrics using the Hadoop APIs, and rely on that dependency to do the rest. As Accumulo developers, we're not necessarily experts on configuring the possible sinks or what happens to the metrics data after. I've actually found configuring metrics to be very confusing and error prone, when I've done it for testing (the conversation you referenced was one such occasion where I found a sort of half solution, just to get it to work well enough for the testing I was doing at the time). My suggestion would be to reach out to our user list instead of the developer list, in case other users have experience with those components in their Accumulo deployments. Or, you could seek assistance directly from the developers for the relevant metrics component, whether it's from the Hadoop team for the GraphiteSink, InfluxDB's developers, or another component's developers. Sorry, I don't mean to pass the buck here, just trying to manage expectations based on the kind of expertise you can expect on this list, and suggest alternative resources in case you don't find an answer here. If you do find a solution, feel free to propose an updated blog post or documentation update for the website. On Thu, Sep 9, 2021, 12:59 Ligade, Shailesh [USA] <[email protected]> wrote: > Thanks for reply, > > > 1. I am using influxdb 1.8.1 and Grafana 7.6 the blog post is using > older versions. - InfluxDB v0.9.4.2 and Grafana v2.5.0. > The hadoop-metrics2-accumulo.properties has the same content as from the > blog > > *.period=30 > accumulo.sink.graphite.class=org.apache.hadoop.metrics2.sink.GraphiteSink > accumulo.sink.graphite.server_host=<INFLUXDB_HOST> > accumulo.sink.graphite.server_port=2003 > accumulo.sink.graphite.metrics_prefix=accumulo > If I setup file sink it is working as well, > > As far as measurements in influxdb are concerned are very different names.. > > e.g. the dashboard is looking for measurement > 'accumulo_tserver_general_IngestRate' and the measurement I can see is > accumulo.tserver.general.Context=tserver.ProcessName=TabletServer.Hostname=<tserverhostname>.ingestRate > and that measurement has no data...not sure how to fix that. > > -S >
