Bug 1594491 - current 3.10 puddle build deploys docker-registry PV twice when using CNS
Summary: current 3.10 puddle build deploys docker-registry PV twice when using CNS
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: Red Hat Gluster Storage
Classification: Red Hat Storage
Component: cns-ansible
Version: rhgs-3.0
Hardware: Unspecified
OS: Unspecified
unspecified
high
Target Milestone: ---
: ---
Assignee: Jose A. Rivera
QA Contact: Ashmitha Ambastha
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2018-06-23 17:18 UTC by Nicholas Schuetz
Modified: 2018-08-07 15:13 UTC (History)
15 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-08-07 15:11:02 UTC
Embargoed:


Attachments (Terms of Use)
ansible inventory file (12.49 KB, text/plain)
2018-06-23 17:18 UTC, Nicholas Schuetz
no flags Details
Inventory file for latest builds and registry (6.86 KB, text/plain)
2018-07-19 12:04 UTC, Ashmitha Ambastha
no flags Details

Description Nicholas Schuetz 2018-06-23 17:18:42 UTC
Created attachment 1453955 [details]
ansible inventory file

Deploying the latest puddle build (3.10.5) with CNS enabled for all infra pods yields two docker-registry PVs of equal size.

pvc-46ce6d13-766c-11e8-ac7b-525400185802   25Gi       RWX            Delete           Bound       default/registry-claim                      glusterfs-storage                   18h
pvc-95ff382c-766c-11e8-ac7b-525400185802   10Gi       RWO            Delete           Bound       openshift-infra/metrics-cassandra-1         glusterfs-storage-block             18h
pvc-bee2c666-766c-11e8-ac7b-525400185802   10Gi       RWO            Delete           Bound       openshift-logging/logging-es-0              glusterfs-storage-block             18h
pvc-ed0ba9da-766c-11e8-ac7b-525400185802   10Gi       RWO            Delete           Bound       openshift-metrics/prometheus                glusterfs-storage                   18h
pvc-ed78dcb6-766c-11e8-ac7b-525400185802   10Gi       RWO            Delete           Bound       openshift-metrics/prometheus-alertmanager   glusterfs-storage                   18h
pvc-ede9f936-766c-11e8-ac7b-525400185802   10Gi       RWO            Delete           Bound       openshift-metrics/prometheus-alertbuffer    glusterfs-storage                   18h
registry-volume                            25Gi       RWX            Retain           Available  


resitry-volume is unused.  Attaching ansible inventory file that i used.

Comment 2 Humble Chirammal 2018-07-02 06:53:06 UTC
Jose, Can you please look into this ?

Comment 3 Jose A. Rivera 2018-07-02 12:16:44 UTC
Please provide what version of openshift-ansible was used, the full inventory file, and a complete list of the steps you took for installation that led to this situation.

Comment 4 Nicholas Schuetz 2018-07-03 22:59:32 UTC
I already provided the first two (see attachment and text of BZ).  Additionally, here are the steps (exactly) that i used.

https://github.com/nnachefski/ocpstuff/blob/master/install/install-beta.md#-begin

Comment 5 Nicholas Schuetz 2018-07-07 15:24:21 UTC
just installed v3.10.14 and got the same result.  It looks like the old-style hosted PV is deploying along with a dynamic PV (via glusterfs-storage storageclass).  The dynamic PV is the one being used (which i'm ok with).  As it is now, i just manually delete the extra registry-volume PV when the install is complete.

[root@master03 ~]# oc get pv
NAME                                       CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS      CLAIM                    STORAGECLASS        REASON    AGE
pvc-444326d1-81f9-11e8-b3fa-525400185802   25Gi       RWX            Delete           Bound       default/registry-claim   glusterfs-storage             18s
registry-volume                            25Gi       RWX            Retain           Available                                                          50s

Comment 9 Ashmitha Ambastha 2018-07-12 10:50:21 UTC
Hi Humble, 

In my deployment I didn't see this issue. There is just 1 pv for docker registry of a default size of 5Gi

------------------ oc get pv output ---------------

# oc get pv
NAME                                       CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS    CLAIM                                 STORAGECLASS               REASON    AGE
pvc-485dd2e6-6f44-11e8-be0b-005056a56419   5G         RWO            Delete           Bound     openshift-infra/metrics-cassandra-1   glusterfs-registry-block             28d
pvc-4a1b0746-6f44-11e8-be0b-005056a56419   5G         RWO            Delete           Bound     openshift-infra/metrics-cassandra-2   glusterfs-registry-block             28d
pvc-4bca43f3-6f44-11e8-be0b-005056a56419   5G         RWO            Delete           Bound     openshift-infra/metrics-cassandra-3   glusterfs-registry-block             28d
pvc-8de354f6-6f44-11e8-be0b-005056a56419   10Gi       RWO            Delete           Bound     openshift-logging/logging-es-0        glusterfs-registry-block             28d
pvc-a7a62c13-6f44-11e8-be0b-005056a56419   10Gi       RWO            Delete           Bound     openshift-logging/logging-es-1        glusterfs-registry-block             28d
pvc-c6dd3513-6f44-11e8-be0b-005056a56419   10Gi       RWO            Delete           Bound     openshift-logging/logging-es-2        glusterfs-registry-block             28d
registry-volume                            5Gi        RWX            Retain           Bound     default/registry-claim                                                     28d

--------------------------- End -----------------------------------

Comment 10 Ashmitha Ambastha 2018-07-12 12:12:42 UTC
The oc version in comment #9 is v3.10.15

Comment 11 Humble Chirammal 2018-07-12 12:32:53 UTC
(In reply to Ashmitha Ambastha from comment #10)
> The oc version in comment #9 is v3.10.15

Thanks ! 

Nicholas, it looks like the issue is not reproducible with a latest version. Is it the same case at your end ?

Comment 12 Nicholas Schuetz 2018-07-12 14:37:45 UTC
I'm still getting this issue on 3.10.15.  Can you share with me your CNS settings in the ansible hosts file?

Comment 13 Humble Chirammal 2018-07-18 15:17:38 UTC
(In reply to Nicholas Nachefski from comment #12)
> I'm still getting this issue on 3.10.15.  Can you share with me your CNS
> settings in the ansible hosts file?

Setting needinfo accordingly.

Comment 14 Nicholas Schuetz 2018-07-18 15:21:54 UTC
As stated previously, my hosts file is attached to this BZ.

Comment 15 krishnaram Karthick 2018-07-19 06:31:50 UTC
(In reply to Humble Chirammal from comment #11)
> (In reply to Ashmitha Ambastha from comment #10)
> > The oc version in comment #9 is v3.10.15
> 
> Thanks ! 
> 
> Nicholas, it looks like the issue is not reproducible with a latest version.
> Is it the same case at your end ?

Why is this bug ON_QA? This bug neither has the acks nor the fix. Moving it back to assigned.

Comment 16 Ashmitha Ambastha 2018-07-19 12:04:42 UTC
Created attachment 1460024 [details]
Inventory file for latest builds and registry

Comment 17 Ashmitha Ambastha 2018-07-19 12:28:57 UTC
(In reply to Nicholas Nachefski from comment #12)
> I'm still getting this issue on 3.10.15.  Can you share with me your CNS
> settings in the ansible hosts file?

I've attached an inventory file to deploy the latest OCP, 3.10.18-1, and CNS. The inventory file also deploys metrics, logging and, gluster registry. You should be able to successfully deploy OCP+CNS (latest available builds).

Comment 18 Raghavendra Talur 2018-07-20 11:49:29 UTC
Nicholas,

Please check the inventory file attached and see if it works.

Comment 19 Nicholas Schuetz 2018-07-23 17:29:43 UTC
So i'm using the exact same hosts file, but now with v3.10.23 this bug has been fixed it seems.  I now only get one PV for docker-registry.  So somewhere between v3.10.18-3.10.23 this got fixed.  What was the issue?


[root@master03 ~]# oc get pv
NAME                                       CAPACITY   ACCESS MODES   RECLAIM POLICY   STATUS    CLAIM                                       STORAGECLASS              REASON    AGE
pvc-6a0e5b1b-8e9d-11e8-a1c0-525400185802   10Gi       RWO            Delete           Bound     openshift-infra/metrics-cassandra-1         glusterfs-storage-block             2m
pvc-90508968-8e9d-11e8-a1c0-525400185802   10Gi       RWO            Delete           Bound     openshift-logging/logging-es-0              glusterfs-storage-block             1m
pvc-b8d5de64-8e9d-11e8-a1c0-525400185802   10Gi       RWO            Delete           Bound     openshift-metrics/prometheus                glusterfs-storage                   35s
pvc-b948f9fe-8e9d-11e8-a1c0-525400185802   10Gi       RWO            Delete           Bound     openshift-metrics/prometheus-alertmanager   glusterfs-storage                   31s
pvc-b9bbf0d6-8e9d-11e8-a1c0-525400185802   10Gi       RWO            Delete           Bound     openshift-metrics/prometheus-alertbuffer    glusterfs-storage                   27s
registry-volume                            25Gi       RWX            Retain           Bound     default/registry-claim                                                          5m

Comment 22 Nicholas Schuetz 2018-08-07 15:11:02 UTC
This issue has magically went away with the GA release.  Closing...

Comment 23 Humble Chirammal 2018-08-07 15:13:29 UTC
(In reply to Nicholas Nachefski from comment #22)
> This issue has magically went away with the GA release.  Closing...

We are glad to hear it !!. Thanks for the update Nicholas.


Note You need to log in before you can comment on or make changes to this bug.