diff options
| author |   Gary Wu <gary.i.wu@huawei.com> | 2018-11-13 18:08:02 -0800 |
|---|---|---|
| committer | Gary Wu <gary.i.wu@huawei.com> | 2018-11-13 18:29:01 -0800 |
| commit | efbcaee5e8670b5778b415b88bd9512b5c44d433 (patch) | |
| tree | 69b81d6f6a876d8ffe630d71a8c7c8cc39b114f8 /kubernetes/log/charts | |
| parent | 460e28cbf478ac6df05056cd05bc60468fcf2745 (diff) | |
Revert use of DaemonSet for logstash
Using a DaemonSet for logstash is problematic for HA k8s
because pods are being scheduled on etcd and orchestration
nodes where we do not want them. Reverting back to a
Deployment with a default of 5 replicas instead.
Issue-ID: LOG-181
Change-Id: I8bbc165a6f23c31b372320c3a94181c689f1baef
Signed-off-by: Gary Wu <gary.i.wu@huawei.com>
Diffstat (limited to 'kubernetes/log/charts')
| -rw-r--r-- | kubernetes/log/charts/log-logstash/templates/deployment.yaml | 3 | ||||
| -rw-r--r-- | kubernetes/log/charts/log-logstash/values.yaml | 4 |
2 files changed, 5 insertions, 2 deletions
diff --git a/kubernetes/log/charts/log-logstash/templates/deployment.yaml b/kubernetes/log/charts/log-logstash/templates/deployment.yaml index 86b1a8c264..7f19bc5ea8 100644 --- a/kubernetes/log/charts/log-logstash/templates/deployment.yaml +++ b/kubernetes/log/charts/log-logstash/templates/deployment.yaml @@ -14,7 +14,7 @@ # limitations under the License. apiVersion: extensions/v1beta1 -kind: DaemonSet +kind: Deployment metadata: name: {{ include "common.fullname" . }} namespace: {{ include "common.namespace" . }} @@ -24,6 +24,7 @@ metadata: release: {{ .Release.Name }} heritage: {{ .Release.Service }} spec: + replicas: {{ .Values.replicaCount }} template: metadata: labels: diff --git a/kubernetes/log/charts/log-logstash/values.yaml b/kubernetes/log/charts/log-logstash/values.yaml index 7779ffe490..955a02a11e 100644 --- a/kubernetes/log/charts/log-logstash/values.yaml +++ b/kubernetes/log/charts/log-logstash/values.yaml @@ -39,7 +39,9 @@ config: elasticsearchServiceName: log-es elasticsearchPort: 9200 -# default number of instances matches cluster size via DaemonSet deployment +# default number of instances +# 30+ logs/sec will saturate a single node to 6+ vCores +replicaCount: 5 nodeSelector: {} |