Monitor self-test to see if there are any problems and act accordingly
codefromthecrypt opened this issue · 1 comments
codefromthecrypt commented
Via #150, we are currently downloading zipkin-server via Jitpack. There were some questions about if this is reliable, or if the lazy build of the server notably impacts master performance in a way that matters.
Since the alternative is hacky and verbose, let's just monitor and see if we need to change before using the alternative below:
# This is a parsing alternative to returning the constant:
#
# We are doing this to avoid a delay as snapshots are built lazily in Jitpack,
# and also avoid insecurities about the version built by Jitpack.
zipkin_snapshot_url() {
zipkin_base=https://oss.jfrog.org/oss-snapshot-local/io/zipkin/zipkin-server
zipkin_version=$(curl -sSL $zipkin_base/maven-metadata.xml | sed -n '/<version>/s/.*<version>\([^<]*\)<\/version>.*/\1/p'|tail -1) && \
zipkin_build=$(curl -sSL $zipkin_base/$zipkin_version/maven-metadata.xml | sed -n '/<value>/s/.*<value>\([^<]*\)<\/value>.*/\1/p'| head -1) && \
echo $zipkin_base/$zipkin_version/zipkin-server-$zipkin-exec.jar
}
codefromthecrypt commented
I added output of the /info endpoint into the test script so we can verify if indeed there are hash problems. probably a good idea anyway
+curl --silent localhost:9411/info
2019-09-26 13:37:18.731 INFO 5466 --- [orker-epoll-2-2] c.l.a.i.JavaVersionSpecific : Using version specific APIs from Java9VersionSpecific
{
"zipkin": {
"version": "2.17.1-SNAPSHOT",
"commit": "844c43b"
}
}