66 lines
2.1 KiB
Plaintext
66 lines
2.1 KiB
Plaintext
|
|
||
|
|
# problem with postfix
|
||
|
|
|
||
|
|
1) doesnt populate the mail relay correctly - this needs to do a group check on item[0] or something so we can get mail01 as the actual relay for clients
|
||
|
|
2) sudo nano -cw /etc/postfix/header_check does an outbound replace - we dont want that for alertmanager as we may want to send out different classes of email
|
||
|
|
|
||
|
|
# node exporter stats collection
|
||
|
|
|
||
|
|
systemctl service file /usr/lib/systemd/system/node_exporter.service loads $OPTIONS
|
||
|
|
/etc/sysconfig/node_exporter contains $OPTIONS
|
||
|
|
|
||
|
|
OPTIONS="--collector.textfile.directory /var/lib/node_exporter/textfile_collector"
|
||
|
|
|
||
|
|
Chris used the following switches, and a custom/cut-down dashboard:
|
||
|
|
|
||
|
|
--collector.cpu \
|
||
|
|
--collector.diskstats \
|
||
|
|
--collector.filesystem \
|
||
|
|
--collector.loadavg \
|
||
|
|
--collector.meminfo \
|
||
|
|
--collector.filefd \
|
||
|
|
--collector.netdev \
|
||
|
|
--collector.stat \
|
||
|
|
--collector.netstat \
|
||
|
|
--collector.systemd \
|
||
|
|
--collector.uname \
|
||
|
|
--collector.vmstat \
|
||
|
|
--collector.time \
|
||
|
|
--collector.mdadm \
|
||
|
|
--collector.tcpstat \
|
||
|
|
--collector.hwmon \
|
||
|
|
--collector.arp \
|
||
|
|
--web.max-requests=40 \
|
||
|
|
--web.listen-address=0.0.0.0:{{ node_exporter_port }} \
|
||
|
|
--web.telemetry-path=/metrics
|
||
|
|
|
||
|
|
The full list of node exporter collections is @ https://github.com/prometheus/node_exporter
|
||
|
|
we can disable all collectors and then enable only the once we want
|
||
|
|
|
||
|
|
--web.disable-exporter-metrics
|
||
|
|
arp WANT
|
||
|
|
bonding WANT
|
||
|
|
# boottime - provided in stat
|
||
|
|
cpu WANT
|
||
|
|
diskstats WANT
|
||
|
|
filesystem WANT
|
||
|
|
hardwaremon
|
||
|
|
infiniband WANT
|
||
|
|
loadavgs WANT
|
||
|
|
meminfo WANT
|
||
|
|
# netclass - lots of useful stuff but remains static really
|
||
|
|
netdev WANT
|
||
|
|
nfs WANT
|
||
|
|
nfsd WANT
|
||
|
|
nvme WANT
|
||
|
|
# os - fairly static do we need a million entries of rhel8
|
||
|
|
powersupplyclass
|
||
|
|
sockstat
|
||
|
|
stat WANT
|
||
|
|
thermal_zone
|
||
|
|
time
|
||
|
|
vmstat WANT
|
||
|
|
xfs WANT - we assume nearly everything will run xfs now
|
||
|
|
|
||
|
|
15/~80 collectors should chop out a lot of data
|
||
|
|
we dont know how to marry metric names to fields though? - this guy has a methodology - its nice too https://mac-blog.org.ua/node-exporter-metrics
|