Reputation: 8148
I have the following retention rule set up in my storage-schemas.conf file.
retentions = 1s:5m,1m:1d,15m:90d,1d:10y
My storage aggregation looks like this:
[default]
pattern = .*
xFilesFactor = 0
aggregationMethod = sum
Problem: Any graph rendered from the second retention archive (1m:1d) has Y axis labels at 33% of the correct value. In all other ways the graph is rendered correctly. When I increase the time period to over 24 hours the Y axis labels show correctly.
When I show the same graphs using graphlot instead of graphite the Y axis renders correctly. (except I really prefer the look/feel of graphite to graphlot)
I'm thinking this is a graphite rendering bug. Is there a workaround?
Upvotes: 1
Views: 477
Reputation: 3591
I think that this can happen from several reasons:
1) After you edited the storage aggreagation , did you delete the relevant whisper files? Notice that if you didn't, then it is still working with the default configuration - which means it uses the average method for the aggregation. In this case you'll have to delete your whisper files and check it again.
By the way you can validate each wsp file current aggregation using the whisper-info file: whisper-info.py your_metric_file_name.wsp
2) If you try to get the data points via the render API with JSON format for example, is it works correctly? https://graphite.readthedocs.org/en/latest/render_api.html
If it indeed works correctly then the problem is probably with the web-app rendering. I've bumped with this issue and I solved it by using the minXStep property. See my question + answer over here: Graphite - Multi archives level retrieval issue
Upvotes: 1