cache request miss

Asked by olafkarsten

I run a graphite stack: megacarbon with ceres storage like described in the blogpost:
http://blog.sweetiq.com/2013/01/carbon-daemon-writer-setup-in-megacarbon/

On another server I run collectd with carbon writer plugin. Now I'm trying to display a graph with the submitted values, but I only get one value as soon as I start the carbon server. After that no further values are displayed.

I inspect the network packages on both servers mit ngrep. The packages are send und received. So I think this part works. I have set the carbonlink timeout in local_settings to 60.0 to be sure thats not the problem.

In the cache.log I got this:
Fri Mar 22 09:14:20 2013 :: CarbonLink creating a new socket for ('127.0.0.1', None)
Fri Mar 22 09:14:20 2013 :: CarbonLink sending request for collectd.server.memory-used to ('127.0.0.1', None)
Fri Mar 22 09:15:20 2013 :: Request-Cache miss [d5489509a5d1d79d2f59759d00b37862]
Fri Mar 22 09:15:20 2013 :: Data-Cache miss [094c2a6e437945f2cf106411405a6af4]

I'm not a programmer, so I'm not sure what I should try next. Any hints? Thanks!

Question information

Language:
English Edit question
Status:
Solved
For:
Graphite Edit question
Assignee:
No assignee Edit question
Solved by:
olafkarsten
Solved:
Last query:
Last reply:
Revision history for this message
olafkarsten (post-olaf) said :
#1

from the exception log, if that helps.

Failed CarbonLink query 'collectd.server.memory.memory-used'
Traceback (most recent call last):
  File "/opt/graphite/webapp/graphite/readers.py", line 130, in fetch
    cached_datapoints = CarbonLink.query(self.real_metric_path)
  File "/opt/graphite/webapp/graphite/carbonlink.py", line 91, in query
    results = self.send_request(request)
  File "/opt/graphite/webapp/graphite/carbonlink.py", line 118, in send_request
    result = self.recv_response(conn)
  File "/opt/graphite/webapp/graphite/carbonlink.py", line 131, in recv_response
    len_prefix = recv_exactly(conn, 4)
  File "/opt/graphite/webapp/graphite/carbonlink.py", line 164, in recv_exactly
    data = conn.recv( num_bytes - len(buf) )
timeout: timed out

Revision history for this message
olafkarsten (post-olaf) said :
#2

Maybe this helps. If I delete all folder & files in ceres dir (or whisper - same problem there). The folder structure get regenerated as expected. All the carbon metrics, statsD and collectD metrics structurs are visible in the webapp. But I get only one Datapoint in the graph (yes, yes autoupdate is enabled and i see the cache requests from the webapp auto update in tail -f cache.log). I'm not sure how to test, if the other data just not arrived, or are not saved or not displayed.

Revision history for this message
olafkarsten (post-olaf) said :
#3

okay, installed from scratch a "standard carbon-cache" setup instead of "megacarbon carbon-daemon" and now it works. So for me the problem is solved.