Magnum cluster is not registered in database on failue.

[rajiv]

When manum cluster is created on the environment, where dns is not configured in that case, "magnum cluster-create" does not show any error and respond with cluster ID. But creation of cluster fails and that cluster ID is not registered in database. Correct behaviour must be, if user has received a uuid for the cluster then that uuid must be stored in database. And on failure status of the cluster must be updated.

[rajiv (rajiv-kumar)]

Steps to reproduce.

1) Machine having magnum-conductor service should not be able to resolve hostname therefore there should not be any entry for nameserver in /etc/resolve.conf file.
2) Create cluster with "magnum cluster-create".
3) Now list clusters using "magnum cluster-list".

List returned in step 3 does not have UUID returned in step 2.

[hongbin (hongbin034)]

@rajiv, could you paste your /etc/resolve.conf file?

[rajiv (rajiv-kumar)]

@hongbin,

/etc/resolv.conf file.

# Dynamic resolv.conf(5) file for glibc resolver(3) generated by resolvconf(8)
# DO NOT EDIT THIS FILE BY HAND -- YOUR CHANGES WILL BE OVERWRITTEN
nameserver 192.168.0.2
search openstacklocal

If i add my organization dns server then this issue is not produced.

[Deepak Jon (deepak-kumar-9)]

M also facing same issue

[hongbin (hongbin034)]

Does this patch fix the issue: https://review.openstack.org/#/c/369982/ ? (Sorry, I cannot confirm it by myself since I still cannot reproduce the error.)

[Madhuri Kumari (madhuri-rai07)]

Hi Rajiv,

As stated by Hongbin, https://review.openstack.org/#/c/369982/ fixes your problem.
Be it any issue when cluster-create fails, we should register the failure in Magnum DB. Can you confirm whether above patch fix your issue or not?

For me, the issue was non-supported template. Try with any non-supported template, you can see ClusterNotSupported exception in logs but no entry in db was made. So I made the change where in case of any failure in cluster-create, it is registered in Magnum DB.