diff options
author | Petr Ospalý <p.ospaly@partner.samsung.com> | 2019-08-22 17:33:07 +0200 |
---|---|---|
committer | Petr Ospalý <p.ospaly@partner.samsung.com> | 2019-08-23 12:52:00 +0200 |
commit | 8f834557f7ccaa38b477853b7b1caf7c91784207 (patch) | |
tree | be4645524a0070c8765da483353400a5f39fe27d /ansible/roles/rke/defaults/main.yml | |
parent | 0364bc6ebc6d26877f1db1be342a10b24f51d80b (diff) |
Add configurable etcd storage feature
Plus the option to have etcd on tmpfs (volatile storage) for little
bigger (maybe) speed (NOT FOR PRODUCTION DEPLOYMENT!).
Issue-ID: OOM-2058
Change-Id: I0bbfc4fbae7f6b46e2fae49656437cd6748efd49
Signed-off-by: Petr Ospalý <p.ospaly@partner.samsung.com>
Diffstat (limited to 'ansible/roles/rke/defaults/main.yml')
-rw-r--r-- | ansible/roles/rke/defaults/main.yml | 40 |
1 files changed, 40 insertions, 0 deletions
diff --git a/ansible/roles/rke/defaults/main.yml b/ansible/roles/rke/defaults/main.yml index 2f160fc2..d9c044b6 100644 --- a/ansible/roles/rke/defaults/main.yml +++ b/ansible/roles/rke/defaults/main.yml @@ -6,6 +6,46 @@ kube_config_dir: "{{ ansible_env.HOME }}/.kube" cluster_config_dir: "{{ app_data_path }}/cluster" # Whether dashboard is exposed. rke_dashboard_exposed: true +rke_etcd: + # By default rke creates bind mount: + # /var/lib/etcd -> /var/lib/rancher/etcd + # These parameters provide means of modifying it: + # - custom bind mount + # - option to use volatile storage + + # Custom bind mount + # + # I did not find a proper way (in the docs) how to override the + # defaults so I just abuse the extra_* args for the rke etcd + # service. It means that it will create another mount point in the + # container and you should use different pathnames than default... + # + # The custom bind mount is by default disabled. + enabled_custom_etcd_storage: false + + # Applicated only if custom mount is enabled. + # Paths must be absolute (start with '/') + # + # Path on the kubernetes/etcd node + storage_path: /var/lib/etcd-custom + # Path inside the container where it is mounted. + storage_mountpoint: /var/lib/rancher/etcd-custom + + # On top of it (with or without custom mount) you can use tmpfs + # as a volatile storage. + # + # CAUTION: This will create temporary filesystem (in the memory) + # so if an etcd node will be poweroff then all etcd data will be + # lost!!! + # + # Don't use unless you don't care about your cluster! + # + # This is intended as an attempt to make deployment little bit + # more faster...by default it is disabled. + enabled_unsafe_volatile_storage: false + # Size of the volatile storage - tmpfs (this will eat your RAM) + tmpfs_size: 5G + rke: # rke (rancher) images etcd: rancher/coreos-etcd:v3.2.24-rancher1 |