Kibana 5.4 Dashboard can't display any object and show unknown error, but Visualize is working fine

Hi all,
I create two dashboards and add some objects including Visualize objects and Search objects.
The two dashboards are working fine until yesterday without changing any configurations.

One is for web log which is sent by filebeat through logstash json codec and saved into elasticsearch.
The other one is for system log which is sent by metricbeat and directly saved into elasticsearch.
Now, the two dashboards can't display any object and show unknown error.

I check these objects in Visualize and Discover page, they are all working fine.
The weird thing is, when I delete all objects in a dashboard and add those objects again, some objects can show up, it also get the "Courier Fetch: Cannot read property 'resolve' of null" error message sometimes.


I didn't see any related error in elasticsearch log. (There are some parsing error about logstash save messages into elasticsearch, but I think that is not relevant.)

I'm not sure what the problem is, maybe it's caused by bad performance on single node (4 core, 30GB mem, 12GB mem for elasticsearch) on AWS EC2.
Any advice? Or what else information I should provide?
Thank you very much!

elasticsearch.yml

cluster.name: testelk
node.name: testelk
path.data: /home/bigdata/elk/elasticsearch/data
path.logs: /home/bigdata/elk/elasticsearch/logs
bootstrap.memory_lock: true
network.host: 0.0.0.0
http.port: 9200
discovery.zen.ping.unicast.hosts: ["127.0.0.1"]
discovery.zen.minimum_master_nodes: 1
action.auto_create_index: .security,.monitoring*,.watches,.triggered_watches,.watcher-history*,filebeat*,logstash*,og_log*,metricbeat*

kibana.yml

server.port: 5601
server.host: "xxx.xx.x.xx"
server.name: "testelk"
elasticsearch.url: "http://localhost:9200"

The "Courier" is how Kibana communicates with Elasticsearch, and that's where the error seems to be coming from. Looks like something is expecting a promise, but getting null instead.

If you know how to use your browser's debugger, you can look in the network tab and see if the response from Elasticsearch is successful. My guess is that it is working, since you don't see errors in the logs, but it's possible that the response can't be parsed for some reason. Sharing the response might be helpful, if it doesn't contain sensitive information.

If you don't see an error in the network request, you could try to chase the error down in the code. My guess is the error is coming from the fetch_these code, and is actually happening elsewhere but being reported here. If you are versed in debugging tools, you could try dropping a breakpoint in there and seeing if you can chase it backwards. We're probably not going to be able to reproduce it without the same dataset you have.

Unfortunately, the stack trace isn't very useful here.

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.