Add a CockroachDB PetSet example #28446

tbg · 2016-07-04T03:10:50Z

The example starts a simple five-node cluster with otherwise
default settings (in particular, 3x replication).

This change is

k8s-bot · 2016-07-04T03:11:11Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

k8s-bot · 2016-07-04T03:11:29Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

k8s-bot · 2016-07-04T03:12:17Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

k8s-bot · 2016-07-04T03:12:35Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

k8s-bot · 2016-07-04T03:12:53Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

bprashanth · 2016-07-04T04:58:17Z

examples/cockroachdb/README.md

+
+When all (or enough) nodes are up, simulate a failure like this:
+
+```shell


please provide an example of how to set/get a key:value. An example displaying the replication properties would be nice too (i.e set in 1 read from 5), as would a scale up/down example (kubectl edit the petset and modify replicas to 6, does the new replica catch up to the cluster?). Is there a "master" we should avoid killing during upgrade?

Optionally, you could write something, bounce the whole cluster (delete pods --all), read the thing.
Also mentioning read/write throughput and how that scales with replicas will help.

Done. I didn't go all out because bash isn't the right language for these tests (and we already have them elsewhere).
Also not too much about tradeoffs etc - I think for the toy example it's fine to assume that people generally have an idea of what they're toying with.

bprashanth · 2016-07-04T04:59:19Z

@tschottdorf thanks! @kubernetes/examples @smarterclayton

tbg · 2016-07-04T23:02:28Z

Updated, PTAL.

bprashanth · 2016-07-27T22:02:46Z

examples/cockroachdb/cockroachdb-petset.yaml

+            # We will need some version of `cockroach init` back for this to
+            # work. For now, just do the same in a shell snippet.
+            # Of course this isn't without danger - if node0 loses its data,
+            # it will upon restarting it will simply bootstrap a new cluster


you have "it will" twice.

bprashanth · 2016-07-27T22:03:46Z

ok to test

bprashanth · 2016-07-27T22:04:43Z

@tschottdorf thanks for the example. I've marked it lgtm with just one nit. If you get to it before it goes in, great, if not we can fix it with a follow up.

sorry for the delay, got sidetracked.

tbg · 2016-07-27T22:30:55Z

No worries, addressed your comment, rebased, and squashed.

On Wed, Jul 27, 2016 at 6:05 PM Prashanth B notifications@github.com
wrote:

@tschottdorf https://github.com/tschottdorf thanks for the example.
I've marked it lgtm with just one nit. If you get to it before it goes in,
great, if not we can fix it with a follow up.

sorry for the delay, got sidetracked.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
#28446 (comment),
or mute the thread
https://github.com/notifications/unsubscribe-auth/AE135A_M89sLlyshpa0CIODxFiFGEa-vks5qZ9YsgaJpZM4JEBSZ
.

bprashanth · 2016-07-27T22:32:54Z

Please run:
./hack/verify-boilerplate.sh
and do as directed, I think jenkins verification failed on it (copyright might need fixing)

tbg · 2016-07-27T22:33:17Z

Yep, thanks. Will fix.

tbg · 2016-07-27T22:36:57Z

Done.

a-robinson · 2016-08-18T17:42:03Z

@bprashanth, I have a couple questions around what it would take to flesh this out a little more once @tschottdorf has time to fix the doc munger's complaints:

Is there a good way to use secrets with petsets yet? Meaning if I have a certificate pre-created for each pet (i.e. cert-0, cert-1, etc.) and want the cert to end up with the pet that has the same index, is there a way to do that other than the pet manually grabbing it from the API when it starts up?
If we want clients of cockroach to spread their load among the pets (which are symmetric from the clients' point of view), I assume we'd have to create a second service with a ClusterIP? My understanding is that the headless service just returns a list endpoints, so we'd be depending on clients to properly randomize their usage of them if we relied on just the headless service.

bprashanth · 2016-08-18T18:00:30Z

Secrets aren't a part of identity, we discussed it initially but there are other ways to distribute secrets (i.e generate something like a csr with your own hostname in the init container and hang on some pki service to sign it), and we don't autoprovision secrets like volumes so it's not something petset currently handles.
The governing service is headless meaning it doesn't have clusterIP. You can lookup the name, and it'll return ips, the ordering of those ips will vary with the client (last I checked nslookup randomized, g's net.LookupHost didn't). If you want loadbalancing you need a service or ingress in front.

The example starts a simple five-node cluster with otherwise default setting (in particular, 3x replication).

tbg · 2016-08-18T20:00:09Z

Munged the docs and pushed.

k8s-bot · 2016-08-19T21:44:22Z

Can one of the admins verify that this patch is reasonable to test? If so, please reply "ok to test".
(Note: "add to whitelist" is no longer supported. Please update configurations in kubernetes/test-infra/jenkins/job-configs/kubernetes-jenkins-pull instead.)

This message may repeat a few times in short succession due to jenkinsci/ghprb-plugin#292. Sorry.

Otherwise, if this message is too spammy, please complain to ixdy.

bprashanth · 2016-08-19T21:46:06Z

ok to test

k8s-bot · 2016-08-19T23:02:32Z

GCE e2e build/test passed for commit 6889f83.

k8s-github-robot · 2016-08-20T00:22:19Z

Automatic merge from submit-queue

@bprashanth

Automatic merge from submit-queue Productionize the cockroachdb example a little more Includes: * A service for clients to use * Readiness/liveness probes * An extended graceful termination period * Automatic prometheus monitoring (when prometheus is configured to watch for annotations on services, as in [CoreOS's recent blog post](https://coreos.com/blog/prometheus-and-kubernetes-up-and-running.html), for example) I'm leaving the management of certs to future work, but if anyone that sees this needs help with them in the meantime, don't hesitate to reach out. Successor to #28446 @bprashanth - if you're still interested in / open to an e2e test (as mentioned in cockroachdb/cockroach#5967 (comment)), let me know and I'll put one together. If so, I assume you'd want it as part of the `petset` test group rather than the `examples` tests? cc @tschottdorf **Release note**: ```release-note NONE ```

googlebot added the cla: yes label Jul 4, 2016

tbg force-pushed the example-petset-cockroachdb branch from 2cf8412 to a402b08 Compare July 4, 2016 03:11

tbg mentioned this pull request Jul 4, 2016

prod: mechanisms are needed to allow cockroach to be deployed in Kubernetes cockroachdb/cockroach#5967

Closed

k8s-github-robot assigned zmerlynn Jul 4, 2016

k8s-github-robot added size/L Denotes a PR that changes 100-499 lines, ignoring generated files. release-note-label-needed labels Jul 4, 2016

bprashanth assigned bprashanth and unassigned zmerlynn Jul 4, 2016

bprashanth reviewed Jul 4, 2016
View reviewed changes

tbg force-pushed the example-petset-cockroachdb branch 2 times, most recently from 9c62353 to 16fef39 Compare July 5, 2016 03:14

bprashanth reviewed Jul 27, 2016
View reviewed changes

bprashanth added release-note-none Denotes a PR that doesn't merit a release note. lgtm "Looks good to me", indicates that a PR is ready to be merged. and removed release-note-label-needed labels Jul 27, 2016

tbg force-pushed the example-petset-cockroachdb branch from 16fef39 to 8a55558 Compare July 27, 2016 22:32

tbg mentioned this pull request Jul 27, 2016

docs: talk about kubernetes cockroachdb/cockroach#8082

Closed

tbg force-pushed the example-petset-cockroachdb branch from 8a55558 to 51ee5af Compare July 27, 2016 22:36

bprashanth added lgtm "Looks good to me", indicates that a PR is ready to be merged. and removed lgtm "Looks good to me", indicates that a PR is ready to be merged. labels Jul 27, 2016

k8s-github-robot added do-not-merge DEPRECATED. Indicates that a PR should not merge. Label can only be manually applied/removed. retest-not-required-docs-only and removed retest-not-required-docs-only labels Jul 28, 2016

tbg mentioned this pull request Jul 28, 2016

deploy: CockroachDB on Kubernetes cockroachdb/docs#506

Closed

wojtek-t removed the do-not-merge DEPRECATED. Indicates that a PR should not merge. Label can only be manually applied/removed. label Jul 31, 2016

Add a CockroachDB PetSet example

6889f83

The example starts a simple five-node cluster with otherwise default setting (in particular, 3x replication).

tbg force-pushed the example-petset-cockroachdb branch from 51ee5af to 6889f83 Compare August 18, 2016 20:00

k8s-github-robot removed the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Aug 18, 2016

bprashanth added the lgtm "Looks good to me", indicates that a PR is ready to be merged. label Aug 18, 2016

bprashanth added the retest-not-required label Aug 19, 2016

k8s-github-robot merged commit e4730f7 into kubernetes:master Aug 20, 2016

a-robinson mentioned this pull request Aug 22, 2016

Productionize the cockroachdb example a little more #31125

Merged

tbg deleted the example-petset-cockroachdb branch October 6, 2016 22:08

This was referenced Mar 6, 2017

Secure deployment on Kubernetes cockroachdb/docs#875

Closed

Orchestrate a secure CockroachDB cluster with Kubernetes cockroachdb/cockroach#13857

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add a CockroachDB PetSet example #28446

Add a CockroachDB PetSet example #28446

tbg commented Jul 4, 2016 •

edited by k8s-oncall

Loading

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

bprashanth Jul 4, 2016

tbg Jul 4, 2016

bprashanth commented Jul 4, 2016

tbg commented Jul 4, 2016

bprashanth Jul 27, 2016

bprashanth commented Jul 27, 2016

bprashanth commented Jul 27, 2016

tbg commented Jul 27, 2016

bprashanth commented Jul 27, 2016

tbg commented Jul 27, 2016

tbg commented Jul 27, 2016

a-robinson commented Aug 18, 2016

bprashanth commented Aug 18, 2016

tbg commented Aug 18, 2016

k8s-bot commented Aug 19, 2016

bprashanth commented Aug 19, 2016

k8s-bot commented Aug 19, 2016

k8s-github-robot commented Aug 20, 2016


		When all (or enough) nodes are up, simulate a failure like this:

		```shell

Add a CockroachDB PetSet example #28446

Add a CockroachDB PetSet example #28446

Conversation

tbg commented Jul 4, 2016 • edited by k8s-oncall Loading

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

k8s-bot commented Jul 4, 2016

bprashanth Jul 4, 2016

Choose a reason for hiding this comment

tbg Jul 4, 2016

Choose a reason for hiding this comment

bprashanth commented Jul 4, 2016

tbg commented Jul 4, 2016

bprashanth Jul 27, 2016

Choose a reason for hiding this comment

bprashanth commented Jul 27, 2016

bprashanth commented Jul 27, 2016

tbg commented Jul 27, 2016

bprashanth commented Jul 27, 2016

tbg commented Jul 27, 2016

tbg commented Jul 27, 2016

a-robinson commented Aug 18, 2016

bprashanth commented Aug 18, 2016

tbg commented Aug 18, 2016

k8s-bot commented Aug 19, 2016

bprashanth commented Aug 19, 2016

k8s-bot commented Aug 19, 2016

k8s-github-robot commented Aug 20, 2016

tbg commented Jul 4, 2016 •

edited by k8s-oncall

Loading