graphite-dev team mailing list archive

[Sebastian YEPES <question224546@xxxxxxxxxxxxxxxxxxxxx>]

Question #224546 on Graphite changed:
https://answers.launchpad.net/graphite/+question/224546

Description changed to:
Hello,

I have recently detected some disk IOPS spikes “Write/Read IOPS” and I
am trying to understand what is causing them, if we look closely at the
“Carbon Cache Perf Stats” graphs the write IO spikes start when the
cache size and queues drop. We can also see the system memory and
received metric are stable.

Last 3hs: http://oi46.tinypic.com/2rdjghx.jpg
Past Week: http://oi46.tinypic.com/oiztc8.jpg

Questions:
- Does anyone know what could be causing these disk write IO spikes?
- Can someone please explain me in more detail how the Carbon cache/queue subsystem works
- Are there any recommended settings to better optimize the caching


These are my settings:
[cache]
MAX_CACHE_SIZE = inf
MAX_UPDATES_PER_SECOND = 600
MAX_UPDATES_PER_SECOND_ON_SHUTDOWN = 2000
MAX_CREATES_PER_MINUTE = inf
(The rest of the setting are the defaults)

Update:
 I have also tried setting the MAX_CACHE_SIZE to 50000 and 134217728 but if a use any other setting then "inf", the carbon-cache process starts using 100% of the CPU and some metrics are not displayed.
 Does anyone know what are the value units of the option MAX_CACHE_SIZE -> MB, KB?

Best regards and thanks in advance for any suggestions,
Sebastian

-- 
You received this question notification because you are a member of
graphite-dev, which is an answer contact for Graphite.