Re: Using functions in the graphite web-api of pmwebd
"Nathan Scott" <[email protected]> Fri, 3 Feb 2017 04:41:34 -0500 (EST)
| Newsgroups | gmane.comp.sysutils.pcp |
|---|---|
| Message-ID | <[email protected]> |
Hi Frits, ----- Original Message ----- > Hi everybody, > > I tried using grafana (both the version that comes with the pcp rpm’s and the > latest version) to create some dashboards of PCP archives. When doing this, > you quite quickly run into the need to do conversions (huge pages are > calculated in pages, whilst other memory components are calculated in Kb, > which is how /proc/meminfo externalises this information, as an example). > FWIW, that's a bug - hugepages stats not being available in correct units - I'll fix that up (we'll need to add some new metrics with appropriate units alongside the existing ones I guess). > However, whenever doing so, grafana indicates there is no data to display. > Upon investigating further, it turns out this is because grafana uses the > graphite web-api, which, when for example a simple ‘alias()’ function is > added to the request, simply does not return anything. To make sure it’s not > grafana, but really the graphite interface, I used the graphite interface > and did the same (added the alias() function), which gives exactly the > result: no data/response. > > Is this intended behaviour? Indeed There seems to be alot of cases of missing/sub-optimal error handling in the web API(s) unfortunately. You might find something was logged in pmwed.log, which usually resides in /var/log/pcp/pmwebd/ (depending on platform). > http://pcp.io/man/man3/pmwebapi.3.html#lbAN > <http://pcp.io/man/man3/pmwebapi.3.html#lbAN> clearly says ‘emulate a subset > of the graphite web-api’, however it costed me quite some time to get to the > bottom of this, mainly because of the absence of an error; a response > telling it is not supported would be much more clear. +1 From a quick browse of the code, there seems to be no code to support "alias" in pmwebd. > If that is intended behaviour, it would mean doing even simple investigations > is severely constrained. What would be the other options of simple > (historical) investigations of pcp archives? Currently it seems I first need > to import them in any of the database backends that grafana supports? If non-web interfaces are an option for you, pmchart and the console tools like pmlogsummary, pmdiff, etc are good options. cheers. -- Nathan -=-=-=-=-=-=-=-=-=-=-=- Groups.io Links: You receive all messages sent to this group. View/Reply Online (#15062): https://groups.io/g/pcp/message/15062 View All Messages In Topic (2): https://groups.io/g/pcp/topic/4299457 Mute This Topic: https://groups.io/mt/4299457?uid=174580 New Topic: https://groups.io/g/pcp/post -=-=- pcp mailing list [email protected] https://groups.io/g/pcp/messages -=-=- Change Your Subscription: https://groups.io/g/pcp/editsub?uid=174580 Group Home: https://groups.io/g/pcp Contact Group Owner: [email protected] Terms of Service: https://groups.io/static/tos Unsubscribe: https://groups.io/g/pcp/leave/354243/563757577/xyzzy -=-=-=-=-=-=-=-=-=-=-=-