Skip to first unread message
Jeff Mesnil
Jul 6, 2017, 9:03:13 AM7/6/17
to MicroProfile
Hi,
This is just a heads up to let you know that I have started to look at the MicroProfile Healthcheck API[1] to provide it in WildFly application server.
For now, I have a prototype[2] to try out some ideas and check that we are passing the TCK.
I've sent some PRs to be able to run the TCK test from our implementation (and provides guidelines[3] from other implementations to do the same).
I already found some interesting ideas and use cases that are not necessarily covered by the spec[4].
As described by the spec[4], it is likely that the server runtime (WildFly in my case but it could be also the case for Swarm or others) should provide some out-of-the-box health check procedures itself in addition to the ones provided by user deployment.
For example, the server runtime could provide JVM checks (if I consume more than 90% of the max memory, it's not healthy; if it's excessively GC'ing it's not healthy, etc.).
As it stands now, the TCK expects to get only health check procedures from the user deployment which contradics the spec.
I have also some questions with naming the health check procedures (are they unique across the process?) and other things.
This is just a heads up to let you know that I have started to look at the MicroProfile Healthcheck API[1] to provide it in WildFly application server.
For now, I have a prototype[2] to try out some ideas and check that we are passing the TCK.
I've sent some PRs to be able to run the TCK test from our implementation (and provides guidelines[3] from other implementations to do the same).
I already found some interesting ideas and use cases that are not necessarily covered by the spec[4].
As described by the spec[4], it is likely that the server runtime (WildFly in my case but it could be also the case for Swarm or others) should provide some out-of-the-box health check procedures itself in addition to the ones provided by user deployment.
For example, the server runtime could provide JVM checks (if I consume more than 90% of the max memory, it's not healthy; if it's excessively GC'ing it's not healthy, etc.).
As it stands now, the TCK expects to get only health check procedures from the user deployment which contradics the spec.
I have also some questions with naming the health check procedures (are they unique across the process?) and other things.
I'll open issue and PRs to address them but I just wanted to let you know that the WildFly community is interested by the MicroProfile Healthcheck :)
jeff
jeff
Kevin Sutter
Jul 6, 2017, 3:10:43 PM7/6/17
to MicroProfile, Heiko Braun
Thanks, Jeff, for your interest in getting the Health Check component working with WildFly. The lead for Health Check is Heiko Braun. It looks like you are working the system correctly with your PR's and Issue generation. Thanks for the help!
-- Kevin
-- Kevin
Reply all
Reply to author
Forward
0 new messages