Elasticsearch Cluster 5.5.1-rancher1 not deployable

Question

Elasticsearch Cluster 5.5.1-rancher1 not deployable

ckauf opened this issue 6 years ago · comments

Hi *,

when trying to deploy the ES community template 5.5.1-rancher1 (and also 5.4.0-rancher1 and 5.4.2-rancher1) I get the following error:

Error (can't have more than one primary service [es-client es-data] referencing the same sidekick [es-storage]

I guess this is the problem in the compose file:

es-data:
  [...]
  volumes_from:
    - es-storage

es-client:
  [...]
  volumes_from:
    - es-storage

I am running this software stack:

Component	Version
Rancher	v1.6.15
Cattle	v0.183.45
User Interface	v1.6.42
Rancher CLI	v0.6.8
Rancher Compose	v0.12.5

I have another rancher cluster where the same template can be deployed just fine.
Software Versions:

Component	Version
Rancher	v1.6.14
Cattle	v0.183.37
User Interface	v1.6.37
Rancher CLI	v0.6.7
Rancher Compose	v0.12.5

I've checked the changelog, but could not find any hints what may explain this issue.
Is there anything I can add?

Regards,
Chris

mkoczorowski commented 6 years ago

+1

Krisan Timur · Answer 1 · Wed Mar 28 2018 18:08:10 GMT+0800 (China Standard Time)

Any workaround?

Chris · Answer 2 · Wed Mar 28 2018 18:36:42 GMT+0800 (China Standard Time)

I downgraded my Rancher Cluster to 1.6.14 as described in the section "Important - Upgrade" that can be found here. Unfortunately I do not know enough about the rancher-compose format to fix this, yet.

Jeff Kolb · Answer 3 · Wed Apr 04 2018 00:24:44 GMT+0800 (China Standard Time)

I fixed this by tweaking a few things on the latest version of elasticsearch in my company published public catalog: https://github.com/INTLFCStone/Rancher-Catalog

Take a look see @krisanalfa @ckauf

Joachim Friberg · Answer 4 · Wed Apr 18 2018 16:00:51 GMT+0800 (China Standard Time)

Same issue for version 6.2.3

Vignesh Shankar · Answer 5 · Tue Apr 24 2018 03:08:34 GMT+0800 (China Standard Time)

+1 on the issue. still there on 6.2.3

phynias · Answer 6 · Wed May 02 2018 01:43:33 GMT+0800 (China Standard Time)

+1 is there a fix or workaround ??

Raúl Sánchez · Answer 7 · Fri May 18 2018 22:20:32 GMT+0800 (China Standard Time)

Hi all,

issue is related to #10563 and merged PR #143 included at rancher v1.6.15. It avoid the use same sidekick for distinct services, like es-storage is defined at es-cluster catalog package.

The solution is redefine storage sidekick es-storage to single storage sidekick for every es-cluster service, es-master-storage, es-data-storage and es-client-storage. Created and merged PR #789 that address it.

es-cluster versions should be filtered correctly for rancher version. 5.5.1-rancher2 and 6.2.3-rancher2 would be deployed at rancher v1.6.15 and higher.

Raúl Sánchez · Answer 8 · Fri May 18 2018 22:33:49 GMT+0800 (China Standard Time)

Feel free to reopen if still have issues.