[tor-project] minutes from the sysadmin meeting

Antoine Beaupré anarcat at torproject.org
Mon Feb 3 18:58:10 UTC 2020

Hi everyone!

Here are the usual minutes from the last sysadmin meeting.

# Roll call: who's there and emergencies

anarcat, gaba, hiro, linus and weasel present

# What has everyone been up to

## anarcat

 * worked on evaluating automated install solutions since we'd
   possibly have to setup multiple machines if the donation comes
 * setup new ganeti node in the cluster, fsn-node-03, [#32937][]
 * dealt with disk problems with said ganeti node [#33098][]
 * switched our install process to setup-storage(8) to standardize
   disk formatting in our install automation work [#31239][]
 * decom'd a ARM build box that was having trouble at scaleway
   [#33001][], future of other scaleway boxes uncertain, delegated
   to weasel
 * looked at the test Discourse instance hiro setup
 * new RT queue ("training") for the community folks [#32981][]
 * upgraded meronense to buster [#32998][] surprisingly tricky
 * started evaluating the remaining work for the buster upgrade and
   contacting teams
 * established first draft of a [sysadmin roadmap][] with hiro and
 * worked on a draft "support policy" with hiro [#31243][]
 * deployed (locally) a [Trac batch client][] to create tickets for
   said roadmap
 * sent and received feedback requests
 * other daily upkeed included scaleway/ARM boxes problems, disk usage
   warnings, security upgrades, code reviews, RT queue config and
   debug [#32981][], package install [#33068][], proper headings
   in wiki [#32985][], ticket review, access control (irl in
   [#32999][], old role in [#32787][], key problems), logging issues
   on archive-01 [#32827][], cleanup old rc.local cruft
   [#33015][], puppet code review [#33027][]

[#33027]: https://bugs.torproject.org/33027
[#33015]: https://bugs.torproject.org/33015
[#32827]: https://bugs.torproject.org/32827
[#32787]: https://bugs.torproject.org/32787
[#32999]: https://bugs.torproject.org/32999
[#32985]: https://bugs.torproject.org/32985
[#33068]: https://bugs.torproject.org/33068
[#32998]: https://bugs.torproject.org/32998
[#32981]: https://bugs.torproject.org/32981
[#33001]: https://bugs.torproject.org/33001
[#33098]: https://bugs.torproject.org/33098
[#32937]: https://bugs.torproject.org/32937
[Trac batch client]: https://help.torproject.org/tsa/howto/trac/
[sysadmin roadmap]: https://help.torproject.org/tsa/roadmap/2020/

## hiro

 * Run system updates (probably twice)
 * Documenting install process workflow visually on [#32902][]
 * Handled request from GR [#32862][])
 * Worked on prometheus blackbox exporter [#33027][]
 * Looked at the test Discourse instance
 * Talked to discourse people about using discourse for our blog
 * Preparing to migrate the blog to static [#33115][]
 * worked on a draft "support policy" with anarcat [#31243][]
 * working on a draft policy regarding services [#33108][]

[#33108]: https://bugs.torproject.org/33108
[#33115]: https://bugs.torproject.org/33115
[#32862]: https://bugs.torproject.org/32862
[#32902]: https://bugs.torproject.org/32902

## weasel

 * build-arm-10 is now building arm64 binaries.  We build arm32
   binaries on the scaleway host in paris still.

# What we're up to next

Note that we're adopting a roadmap in this meeting which should be
merged with this step, once we have agreed on the process. So this
step might change in the next meetings, but let's keep it this way for

## anarcat

I pivoting towards stabilisation work and postponed all R&D and other


 * new gnt-fsn node (fsn-node-04) -118EUR=+40EUR [#33081][]
 * unifolium decom (after storm), 5 VMs to migrate, [#33085][]
 * buster upgrade 70% done: 53 buster (+5), 23 stretch (-5)
 * automate upgrades: enable unattended-upgrades fleet-wide

[#33085]: https://bugs.torproject.org/33085
[#33081]: https://bugs.torproject.org/33081


 * install automation tests and refactoring [#31239][]
 * SLA discussion (see below, [#31243][])

[#31243]: https://bugs.torproject.org/31243
[#32462]: https://bugs.torproject.org/32462
[#32351]: https://bugs.torproject.org/32351
[#31239]: https://bugs.torproject.org/31239
[#29387]: https://bugs.torproject.org/29387


 * kvm4 decom [#32802][]
 * varnish -> nginx conversion [#32462][]
 * review cipher suites [#32351][]
 * publish our puppet source code [#29387][]
 * followup on SVN shutdown, only corp missing [#17202][]
 * audit of the other installers for ping/ACL issue [#31781][]
 * email services R&D [#30608][]
 * send root@ emails to RT [#31242][]
 * continue prometheus module merges

[#32802]: https://bugs.torproject.org/33082
[#17202]: https://bugs.torproject.org/17202
[#31781]: https://bugs.torproject.org/31781
[#30608]: https://bugs.torproject.org/30608
[#31242]: https://bugs.torproject.org/31242

## Hiro
 * storm shutdown [#32390][]
 * enable needrestart fleet-wide [#31957][]
 * review website build errors [#32996][]
 * migrate gitlab-01 to a new VM (gitlab-02) and use the omnibus
   package instead of ansible [#32949][]
 * migrate CRM machines to gnt and test with Giant Rabbit [#32198][]
 * prometheus blackbox exporter [#33027][]

[#32390]: https://bugs.torproject.org/32390
[#32198]: https://bugs.torproject.org/32198
[#32949]: https://bugs.torproject.org/32949
[#32996]: https://bugs.torproject.org/32996
[#31957]: https://bugs.torproject.org/31957

# Roadmap review

Review the roadmap and estimates.

We agreed to use trac for roadmapping for february and march but keep
the wiki for soft estimates and longer-term goals for now, until we
know what happens with gitlab and so on.

Useful references:

 * temporal pad where we are sorting out roadmap: https://pad.riseup.net/p/CYOUx21kpxLL_5Eui61J-tpa-roadmap-2020
 * tickets marked for february and march: https://trac.torproject.org/projects/tor/wiki/org/teams/SysadminTeam

# TPA-RFC-1: RFC process

One of the interesting takeaways I got from reading the [guide to
distributed teams][] was the idea of using [technical RFCs as a
management tool][].

 [guide to distributed teams]: https://increment.com/teams/a-guide-to-distributed-teams/
 [technical RFCs as a management tool]: https://buriti.ca/6-lessons-i-learned-while-implementing-technical-rfcs-as-a-management-tool-34687dbf46cb

They propose using a formal proposal process for complex questions

 * might impact more than one system
 * define a contract between clients or other team members
 * add or replace tools or languages to the stack
 * build or rewrite something from scratch

They propose the process as a proposal with minimum of two days and a
maximum of a week discussion delay.

In the team this could take many forms, but what I would suggest would
be a text proposal that would be a (currently Trac) ticket with a
special tag, which would also be explicitely forwarded to the "mailing
list" (currently tpa alias) with the `RFC` subject to outline it.

Examples of ideas relevant for process:

 * replacing Munin with grafana and prometheus [#29681][]
 * setting defaut locale to C.UTF-8 [#33042][]
 * using Ganeti as a clustering solution
 * using setup-storage as a disk formatting system
 * setting up a loghost
 * switching from syslog-ng to rsyslog

[#33042]: https://bugs.torproject.org/33042
[#29681]: https://bugs.torproject.org/29681

Counter examples:

 * setting up a new Ganeti node (part of the roadmap)
 * performing security updates (routine)
 * picking a different machine for the new ganeti node (process wasn't
   documented explicitely, we accept honest mistake)

The idea behind this process would be to include people for major
changes so that we don't get into a "hey wait we did what?" situation
later. It would also allow some decisions to be moved outside of
meetings and quicker decisions. But we also understand that people can
make mistakes and might improvise sometimes, especially if something
is not well documented or established as a process in the
documentation. We already have the possibility of doing such changes
right now, but it's unclear how that process works or if it works at
all. This is therefore a formalization of this process.

If we agree on this idea, anarcat will draft a first meta-RFC
documenting this formally in trac and we'd adopt it using itself,
bootstrapping the process.

We agree on the idea, although there are concerns about having too
much text to read through from some people. The first RFC documenting
the process will be submitted for discussion this week.

# TPA-RFC-2: support policies

A *second* RFC would be a formalization of our support policy, as per:

Postponed to the RFC process.

# Other discussions

No other discussions, although we worked more on the roadmap after the
meeting, reassigning tasks, evaluating the monthly capacity, and
estimating tasks.

# Next meeting

March 2nd, same time, 1500UTC (which is 1600CET and 1000EST).

# Metrics of the month

 * hosts in Puppet: 77, LDAP: 80, Prometheus exporters: 124
 * number of apache servers monitored: 32, hits per second: 158
 * number of nginx servers: 2, hits per second: 2, hit ratio: 0.88
 * number of self-hosted nameservers: 5, mail servers: 10
 * pending upgrades: 110, reboots: 0
 * average load: 0.34, memory available: 328.66 GiB/1021.56 GiB,
   running processes: 404
 * bytes sent: 160.29 MB/s, received: 101.79 MB/s
 * completion time of stretch major upgrades: 2020-06-06

Antoine Beaupré
torproject.org system administration
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 487 bytes
Desc: not available
URL: <http://lists.torproject.org/pipermail/tor-project/attachments/20200203/75038089/attachment.sig>

More information about the tor-project mailing list