search

pires / kubernetes-elasticsearch-cluster

Elasticsearch cluster on top of Kubernetes made easy.
Apache License 2.0
1.51k stars 690 forks source link

Indexing failing on _bulk API calls #203

Open syst0m opened 6 years ago

syst0m commented 6 years ago

Getting the following errors in nginx pod, while indexing:

[error] 2979#2979: *2143972 upstream prematurely closed connection while reading response header from upstream, client: 10.132.0.4, server: XXXX, request: "POST /_bulk?timeout=1m HTTP/1.1", upstream: "http://10.40.4.20:9200/_bulk?timeout=1m", host: "XXXXXX" Expand all | Collapse all { So, looks like the client nodes are not responding within the defined timeout. This is set in nginx thusly:


            proxy_connect_timeout                   5s;
            proxy_send_timeout                      60s;
            proxy_read_timeout                      60s;

[error] 47#47: *1759038 user "XXXX" was not found in "/etc/ingress-controller/auth/default-ingress-with-auth.passwd", client: 10.132.0.5, server: XXXX, request: "POST /_bulk?timeout=1m HTTP/1.1", host: "jXXX:80"

User is present and this only fails intermitently. There are no resources related issues. Additionaly, no issues occuring with k8s cluster operation at the times this fails.