11:17 a.m.
Hello
I have tried to run Hawkular in both `oc cluster up` and OpenShift,
different versions of both.
The documentation for `oc cluster up` says (
https://github.com/openshift/origin/blob/master/docs/cluster_up_down.md#i...)
that by using the `--metrics` parameter, metrics is installed.
Its a similar story when installing OpenShift using Openshift-ansible.
However, in both cases, when logging into the OpenShift webconsole I see an
error message saying:
An error occurred getting metrics. Open Metrics URL | Don't Show Me Again
It would seem there is more to correctly installing Hawkular metrics than
just using the parameter.
What else needs to be done? And is it possible get the additional info
added to the documentation for openshift-ansible and `oc cluster up`?
Thanks
12:51 p.m.
----- Original Message -----
What happens if you open the metrics URL?
By default the metrics certificate is self signed and not trusted by the browser. So you
will first need to click on the link and verify if you trust the certificate or not. If
you accept the certificate and refresh the metrics page, do you still see the error?
If not, you might be running into an issue. If that is the case, can you please attach the
output of 'oc get pods -n openshift-infra' to see what state each of the pods are
in?
2:35 p.m.
Hi Matt
On 11 April 2017 at 19:51, Matt Wringe <mwringe(a)redhat.com> wrote:
Its a dns issue, rather than a cert issue:
This site can’t be reached
*hawkular-metrics.apps.console.mydomain.com
<http://hawkular-metrics.apps.console.mydomain.com>*’s server DNS addresscould
not be found.
DNS_PROBE_FINISHED_NXDOMAIN
# oc get pods -n openshift-infra
NAME READY STATUS RESTARTS AGE
hawkular-cassandra-1-qnx2x 1/1 Running 0 3d
hawkular-metrics-3qhpj 1/1 Running 5 3d
heapster-69qzn 1/1 Running 4 3d
2:47 p.m.
I created a new route, and this time I get a different two different errors.
*In http:*
502 Bad GatewayThe server returned an invalid or incomplete response.
*In https:*
Application is not available
The application is currently not serving requests at this endpoint. It may
not have been started or is still starting.
Possible reasons you are seeing this page:
- *The host doesn't exist.* Make sure the hostname was typed correctly
and that a route matching this hostname exists.
- *The host exists, but doesn't have a matching path.* Check if the URL
path was typed correctly and that the route was created using the desired
path.
- *Route and path matches, but all pods are down.* Make sure that the
resources exposed by this route (pods, services, deployment configs, etc)
have at least one pod running.
On 11 April 2017 at 21:35, Anton <kurrent93(a)gmail.com> wrote:
1:55 p.m.
This happens always - and I never quite figured out why. But once you click on that link
"Open Metrics URL" and get the "started" splash page, it all shows it
is working.
I think it has to do with the self-signed certificate and how the browser wants to be told
it is OK. I'm not sure, that's just a guess.
But, if you started "oc cluster up --metrics" and you see that message, its all
OK. Just click "Open Metrics URL" and see that you can get to the metrics splash
screen via your browser.
Other than that, I don't have a good explanation - I just know how I get that message
to go away. It shows up for me everytime I "oc cluster up --metrics".
----- Original Message -----
4:12 p.m.
Hi John
To clarify, its not a certificate issue. After I accept the certificate
warning, the browser redirects to:
Application is not available
The application is currently not serving requests at this endpoint. It may
not have been started or is still starting.
Possible reasons you are seeing this page:
The host doesn't exist. Make sure the hostname was typed correctly and that
a route matching this hostname exists.
The host exists, but doesn't have a matching path. Check if the URL path
was typed correctly and that the route was created using the desired path.
Route and path matches, but all pods are down. Make sure that the resources
exposed by this route (pods, services, deployment configs, etc) have at
least one pod running.
On 11 April 2017 at 20:55, John Mazzitelli <mazz(a)redhat.com> wrote:
4:28 p.m.
(you are going to get this reply about 2 to 3 hours after i send it because the mailing
list seems to delay my posts - I'm sending this a few minutes after I saw your post
though :)
If that message is showing up even after, say, 5 minutes, then it sounds like Origin
Metrics is failing to start.
If you look at your "oc cluster up --metrics" output, see what kind of messages
you get regarding starting the metrics component.
If you look in your openshift-infra project, you should see the metrics pod - any errors
indicated there? In the events tab or logs tab or something like that?
Not sure why metrics wouldn't start, but I do know that I have a old machine where it
takes a long time for the metrics stuff to start up fully (it has to not only download the
image from docker hub the first time, but starting the metrics components can take a while
on my slow machine).
----- Original Message -----
5:06 a.m.
I thought that the problem might be wtih how my domain name is configured,
so I set up a CNAME record to point to the ip of my cluster, but this still
did not work.
It seems something internally is not working internally with regards to the
routing/dns to the hawkular-metrics service.
To eleborate, I created a new route, to hawkular-metrics, and this new
route DOES work.
So I deleted the originally created route, and made a new one, with the
same route name - 'hawkular-metrics' - thinking that any openshift deployed
services would look up the route name. But it seems this is not the case.
It seems it is the route url that is being used.
Is it possible to change the url for hawkular-metrics?
On 11 April 2017 at 23:28, John Mazzitelli <mazz(a)redhat.com> wrote:
2912
days inactive
2913
days old
7 comments
3 participants
participants (3)
-
Anton -
John Mazzitelli -
Matt Wringe