[tech] Tech/Wheel Meeting 2021-01-10 14:00 - One hour reminder
root
root at ucc.gu.uwa.edu.au
Sun Jan 10 13:00:01 AWST 2021
Wheel Meeting Agenda - Sunday 2021-01-10 14:00
==============================================
- VENUE: UCC Clubroom
- and online at https://meetings.ucc.asn.au/b/bob-yrk-uy6 ?
*Meeting opened xx:xx*
## Attendance
- Present
- Apologies
- Absent
## Next meeting
- Schedule next meeting
- what's happening before O-Day 2021-02-19 ?
- ACTION: xxx, who hasn't tried it recently?:
- Update the agenda, update the crontab, check at T-7days that the notice really went out
- Set and verify reminders of next meeting: `motsugo# crontab -e`
- skip the `4day` , unless there's issues at `1week` ?
- Curate agenda.next
## Standing items (brief)
### Visibly reinduct members new (and old?) with the "Wheel Group Ethical Guidelines"
- examining an Ethical Guideline, e.g. asking:
- What's an example situation in which it could be encountered?
- What other guidelines or rules could it conflict with? How would
one resolve it?
### Status check: Regular updates, monitoring
- atlantic cluster
- should we `apt upgrade` ? do some reboots?
- e.g. Debian oldstable 9 "stretch" -> Debian stable 10 "buster"
- find candidates on ocsinventory
- has it stopped reporting versions in Debian 10?
https://ocsinventory.ucc.asn.au/ocsreports/index.php?function=visu_search&fields=HARDWARE-LASTCOME&comp=tall&values=07/07/2020%2007:19&values2=&type_field=
- molmol
- Dead SSD? at Mon 9 Nov 08:00:10 AWST 2020
```
molmol: /space/scratch/nick>zpool status|grep -C4 DEGRADED
logs
mirror-4 DEGRADED 0 0 0
ada0p3 ONLINE 0 0 0
3087349144323640050 UNAVAIL 0 0 0 was /dev/gpt/molmol-slog0
```
- Upgrade and performance analysis:
1. monitoring: add prometheus metric export
2. iozone performance-and-latency-under-load benchmark
3. enable metaslab debugging mode: https://serverfault.com/questions/511154/zfs-performance-do-i-need-to-keep-free-space-in-a-pool-or-a-file-system
4. iozone performance-and-latency-under-load benchmark
5. OS upgrade
6. iozone performance-and-latency-under-load benchmark
### Status check: Backups
### Status check: Password/Key rotations
- https://en.wikipedia.org/wiki/Pro_re_nata
## ..._then_ New wheel members, additions, nominations
- Welcome to wheel!
- Read /home/wheel/docs/WelcomeToWheel
- winadmin, sprocket
- [BRD]@2020-08-13 `uid=12426(bird) gid=10021(gumby) groups=10021(gumby),10069(committee),12203(door),666(winadmin),777(sprocket)`
- `uid=12469(hilmi) gid=10021(gumby) groups=10021(gumby)`
## New Matters
- [NTU] `Dec 8 05:41:46 motsugo kernel: [10920663.319201] Oops: 0000 [#1] SMP PTI`
- thanks for the remote power cycle, [TPG]
- `murasoi:/var/log/ucc/messages` is pretty noisy!
- can we fix some of the root causes of the noise?
- kerosene's timezone is UTC, probably should be local?
- [TRS]@2020-11-03: SOGo ( https://sogo.nu/ ) has been down for a while too
- [NTU] molmol had stopped responding - out of memory and the wrong thing got killed?
- remote power cycle of molmol
- OOM possibly triggered by rdiff-backup on huge files?
- ACTION: [???] clean up the huge files, see `mollitz:/backups/log`
- is SOGo working again?
- ACTION: [???] can we add a prometheus+grafana health check for SOGo?
- [NTU] Live demo: ceph magikarp:osd.3 is smaller than it should be, let's fix it
## Matters arising previously
- ACTION: xxx, who hasn't tried it recently?: Set and verify reminders of next meeting: `motsugo# crontab -e`
- ACTION: DONE? [MTL]+[MPT] poking zonemake.py and its API-driven replacements and children
- ACTION: DONE? [MPT] cf_tools / zonemake.py / octodns: generate API tokens for uccpass
- https://lists.ucc.gu.uwa.edu.au/pipermail/tech/2020-December/005411.html
- [NTU] How can internal-only proxmox cluster VM hosts automate new letsencrypt certs?
- ACTION: [MTL] to look at UCC web reverse proxies
- ACTION: [MPT] UWA IT liason: matrix test domain
- ACTION: [MPT] update https://wiki.ucc.asn.au/Network with latest traffic paths
- ACTION: [TEC] to look at dashboards for murasoi network traffic
- [NTU] can it capture when bulk TCP resets are sent by upstream connection-tracking routers?
- graph the age of existing non-LAN connections, look for spikes to zero?
- graph new connections, look for spikes?
*Meeting closed xx:xx*
----
```
# https://demo.hedgedoc.org/Hlsapf47RsqpgIjqLVfMUw
cd /home/wheel/docs/meetings
CODIMD_SERVER=https://demo.hedgedoc.org codimd export --md Hlsapf47RsqpgIjqLVfMUw ./$(date +%Y-%m-%d).txt
git commit -a "minutes"
```
# vim: tabstop=4 shiftwidth=4 expandtab
More information about the tech
mailing list