Closed Cardds closed 6 years ago
Looks interesting, thank you!
Please provide a good documentation section in the README as well to go along with this and how it can be configured in lieu of the existing behaviors of the SPI. The documentation will need a good update.
Also please update the .travis.yml
config to incorporate appropriate tests for this new config option
Added basic test cases in separate commit. More intensive/expansive tests may be added in a separate PR at a later date. Note that timeout/file length was increased for this section due to inconsistency in reaction times to cluster state change.
README & docs to be added in the next commit.
The README has been updated with comments on the XML configuration.
Thanks! please take a look at: https://travis-ci.org/bitsofinfo/hazelcast-docker-swarm-discovery-spi/builds/332887211
Build failed on the section that hasn't been altered. This also happened on an earlier commit that did not have any modifications to travis yml.
Noticed that hazelcast check/recovery speed isn't determinate in these sets of tests. This has been mitigated somewhat by Increasing the sleep time from 60 to 80 and increasing the number of logs fetched from 50 to 150 in the newly added section. However, this does not eliminate false alarms.
Released and published to bintray/maven, thanks man!
Adds discovery strategy that relies on endpoint mode being set to dnsrr in docker, but eliminates the need for addresses of the swarm manager (and any need for exposing the docker HTTP API port).