Many things have been rescheduled to march and april because we ran out of time to do what we wanted. In particular, the libvirt/kvm migrations are taking more time than expected.
# Policies review
TPA-RFC-1: policy; marked as adopted
TPA-RFC-2; support; hiro to write up a draft.
TPA-RFC-3: tools; to be brainstormed here
The goal of the new RFC is to define which *tools* we use in TPA. This
does not concern service admins, at least not in the short term, but
only sysadmin stuff. "Tools", in this context, are programs we use to
implement a "service". For example, the "mailing list" service is
being ran by the "mailman" tool (but could be implemented with
another). Similarly, the "web cache proxy" service is implemented by
varnish and haproxy, but is being phased out in favor of Varnish.
Another goal is to *limit* the number of tools team members should
know to be functional in the team, and formalize past decisions (like
"we use debian").
We particularly discussed the idea of introducing Fabric as an "ad-hoc
changes tool" to automate host installation, retirement, and
reboots. It's already in use to automate libvirt/ganeti migrations and
is serving us well there.
# Other discussions
A live demo of the Fabric code was performed some time after the
meeting and no one raised objections to the new project.
# Next meeting
No discussed, but should be on april 6th 2020.
# Metrics of the month
* hosts in Puppet: 77, LDAP: 81, Prometheus exporters: 124
* number of apache servers monitored: 31, hits per second: 148
* number of nginx servers: 2, hits per second: 2, hit ratio: 0.89
* number of self-hosted nameservers: 6, mail servers: 10
* pending upgrades: 174, reboots: 0
* average load: 0.63, memory available: 308.91 GiB/1017.79 GiB,
