HTTP request error, and can't fetch metrics

[epvklout]

I've set up a new 0.9.12 graphite server and am using grafana from the 1.0.4 tarball. Graphite has no auth enabled. Graphite and elasticsearch are on the same host, which is directly reachable by the browser.

When grafana starts in the browser, the test graph (randomwalk, etc) shows "Oops! HTTP Request error". tcpdump shows that the browser sends a POST /render/ request containing:

from=-6h&until=now&target=randomWalk('random%20walk')&target=randomWalk('random%20walk2')&target=randomWalk('random%20walk3')&maxDataPoints=600&format=json

and receives back a "HTTP/1.1 200 OK" result containing:

[{"target": "random walk", "datapoints": [[0, 1391610485], [-0.32153186425113056, 1391610545], [-0.7608628977570804, 1391610605], ... ... ... }]

Further, when using Configure Dashboard -> Metrics -> Load All, the browser sends to graphite "GET /metrics/index.json HTTP/1.1" and receives a "HTTP/1.1 200 OK" result back containing:

["carbon.agents.my_graphite_server-a.avgUpdateTime", "carbon.agents.my_graphite_server-a.cache.overflow", "carbon.agents.my_graphite_server-a.cache.queries", "carbon.agents.my_graphite_server-a.cache.queues", "carbon.agents.my_graphite_server-a.cache.size", "carbon.agents.my_graphite_server-a.committedPoints", "carbon.agents.my_graphite_server-a.cpuUsage", "carbon.agents.my_graphite_server-a.creates", "carbon.agents.my_graphite_server-a.errors", "carbon.agents.my_graphite_server-a.memUsage", "carbon.agents.my_graphite_server-a.metricsReceived", "carbon.agents.my_graphite_server-a.pointsPerUpdate", "carbon.agents.my_graphite_server-a.updateOperations", "servers.127_0_0_1_9400.sun_management_GarbageCollectorImpl.CollectionCount", "servers.127_0_0_1_9400.sun_management_GarbageCollectorImpl.CollectionTime",
... ]

and the display in grafana shows a red box containing:

{"data":"","status":0,"config":{"transformRequest":[null],"transformResponse":[null],"method":"GET","url":"http://my_graphite_server/metrics/index.json","headers":{"Accept":"application/json, text/plain, */*"}}} 

I've tried pointing it at a number of other graphite servers with the same results, but they are running 0.9.10 instead of 0.9.12. The only things I have edited in config.js are elasticsearch and graphiteUrl. I've also tried both the 1.0.4 grafana tarball and a copy cloned from master.

I'm not sure what's going wrong here but it does not appear to be failing to talk to the graphite server, since I can watch the network exchange and it looks correct.

[torkelo]

Strange, the randomWalk graphite request returns data and status code 200, grafana somehow gets a 500 error? Very strange. Try other time ranges, and with real metric expressions. (not just randomWalk).

Regarding:
Configure Dashboard -> Metrics -> Load All, this is a feature that is not ready, should have removed this tab in the release but forgot to.

[torkelo]

Are you still having http error issues?