5b6aa3bc95
* k8s: add tolerations for 'nodeType=crawling:NoSchedule' to allow scheduling crawling on designated nodes for crawler and profiles jobs and statefulsets * add affinity for 'nodeType=crawling' on crawling and profile browser statefulsets * refactor crawljob: combine crawl_updater logic into base crawl_job * increment new 'crawlAttemptCount' counter crawlconfig when crawl is started, not necessarily finished, to avoid deleting configs that had attempted but not finished crawls. * better external mongodb support: use MONGO_DB_URL to set custom url directly, otherwise build from username, password and mongo host
102 lines
2.1 KiB
YAML
102 lines
2.1 KiB
YAML
---
|
|
apiVersion: apps/v1
|
|
kind: StatefulSet
|
|
metadata:
|
|
name: browser-{{ id }}
|
|
namespace: {{ namespace }}
|
|
labels:
|
|
browser: {{ id }}
|
|
role: browser
|
|
|
|
spec:
|
|
selector:
|
|
matchLabels:
|
|
browser: {{ id }}
|
|
role: browser
|
|
|
|
serviceName: browser-{{ id }}
|
|
replicas: 1
|
|
|
|
template:
|
|
metadata:
|
|
labels:
|
|
browser: {{ id }}
|
|
role: browser
|
|
|
|
spec:
|
|
affinity:
|
|
nodeAffinity:
|
|
preferredDuringSchedulingIgnoredDuringExecution:
|
|
- weight: 1
|
|
preference:
|
|
matchExpressions:
|
|
- key: nodeType
|
|
operator: In
|
|
values:
|
|
- "{{ crawler_node_type }}"
|
|
|
|
podAffinity:
|
|
preferredDuringSchedulingIgnoredDuringExecution:
|
|
- weight: 2
|
|
podAffinityTerm:
|
|
topologyKey: "failure-domain.beta.kubernetes.io/zone"
|
|
labelSelector:
|
|
matchLabels:
|
|
job-name: job-{{ id }}
|
|
crawl: {{ id }}
|
|
|
|
tolerations:
|
|
- key: "nodeType"
|
|
operator: "Equal"
|
|
value: "crawling"
|
|
effect: "NoSchedule"
|
|
|
|
containers:
|
|
- name: browser
|
|
image: {{ crawler_image }}
|
|
imagePullPolicy: {{ crawler_image_pull_policy }}
|
|
command:
|
|
- create-login-profile
|
|
- --interactive
|
|
- --filename
|
|
- /tmp/profile.tar.gz
|
|
- --url
|
|
- {{ url }}
|
|
{%- if profile_filename %}
|
|
- --profile
|
|
- "@{{ profile_filename }}"
|
|
{%- endif %}
|
|
|
|
envFrom:
|
|
- secretRef:
|
|
name: storage-{{ storage_name }}
|
|
|
|
env:
|
|
- name: STORE_PATH
|
|
value: {{ storage_path }}
|
|
|
|
---
|
|
apiVersion: v1
|
|
kind: Service
|
|
metadata:
|
|
name: browser-{{ id }}
|
|
labels:
|
|
browser: {{ id }}
|
|
role: browser
|
|
|
|
spec:
|
|
clusterIP: None
|
|
selector:
|
|
browser: {{ id }}
|
|
role: browser
|
|
|
|
ports:
|
|
- protocol: TCP
|
|
port: 9223
|
|
name: browser-api
|
|
|
|
- protocol: TCP
|
|
port: 9222
|
|
name: browser-ws
|
|
|