You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
i've enabled PCP hotproc metrics using hotproc.conf (cpuburn > 0.10) & also seup sasl based authentication and installed redis as well as grafana everything on a test server.
the real issue I've observed for most metrics example "hotproc.predicate.cpuburn" shows up instances (40digit alpha-numeric number) who're more than 10% CPU usage as per predicate but no PID and PROC name associated with it. Sometime takes its own time and sometimes it doesn't at all.
my env:
RHEL8.10
PCP 5.3
Grafana 9.2
grafana-pcp 5.1
Redis 5.0
so am I missing anything here?
The text was updated successfully, but these errors were encountered:
I've setup hotproc metrics going into Valkey now (Fedora40, pcp-6.3.1) and everything seems to be fine from the PCP end. There may be some effort needed on the Grafana end, not sure, but this is what I see using just PCP tools (i.e. pmseries command line, which talks directly to Valkey/Redis):
nathans@fedora:~/git/pcp$ pmseries hotproc.psinfo.rss
014f2aaf4d97f588a00c52ea3fcc672fbd423ecb
nathans@fedora:~/git/pcp$ pmseries -i `pmseries hotproc.psinfo.rss`| head
014f2aaf4d97f588a00c52ea3fcc672fbd423ecb
inst [1350 or "001350 /usr/libexec/pcp/bin/pmcd"] series e720ad93a6039f435c0ce497129091b7a59be7c7
inst [1354 or "001354 /var/lib/pcp/pmdas/root/pmdaroot"] series 782aeda842aac7a4a64a103639976f654d418148
inst [1360 or "001360 /var/lib/pcp/pmdas/proc/pmdaproc"] series e1ae808971df1715ae8ba61e3135db85cf53e299
inst [1361 or "001361 /var/lib/pcp/pmdas/xfs/pmdaxfs"] series 2cce339576c439f6f792df35eaf75ba339465892
inst [1362 or "001362 /var/lib/pcp/pmdas/sample/pmdasample"] series 7e1f58318cc56efa0eeca1de410995ab1229c5de
inst [1363 or "001363 /var/lib/pcp/pmdas/statsd/pmdastatsd"] series eff572062f652a3f888239be08bf7156f1c76892
inst [1367 or "001367 /var/lib/pcp/pmdas/linux/pmdalinux"] series 20017f275086c90adfaeb8cdb8d76d4df712f6eb
inst [1371 or "001371 /usr/bin/python3"] series e0632ae864826d8addcb01a8e8718df6d68d0ae2
nathans@fedora:~/git/pcp$
so, the process identifiers and command names (i.e. PCP instances for hotproc metrics), is there - there may be some magic required to display these in Grafana rather than the 40-digit SID (series identifier). @sfeifer or @andreasgerstmayr may be able to help us further there.
i've enabled PCP hotproc metrics using hotproc.conf (cpuburn > 0.10) & also seup sasl based authentication and installed redis as well as grafana everything on a test server.
the real issue I've observed for most metrics example "hotproc.predicate.cpuburn" shows up instances (40digit alpha-numeric number) who're more than 10% CPU usage as per predicate but no PID and PROC name associated with it. Sometime takes its own time and sometimes it doesn't at all.
my env:
so am I missing anything here?
The text was updated successfully, but these errors were encountered: