Some onionperf clients are running out of disk space and we should probably archive their data logs (#40038 (closed)).
@acute suggested we can take this opportunity to bump the onionperf client version and deploy new instances that could coexist for a few weeks (so we do not lose measurements).
Just dropping here the checklist based on deploying the previous instances, please change if anything doesn't look right or if things were missed on the monitoring side!
tag onionperf release 0.10
create 3 new VMs (op-hk7, op-hk7a, op-de7a) on AWS and 2 new VMs on Greenhost (op-us7, op-nl7)
ask the admin team to add DNS entries for the instances
deploy Onionperf 0.10, Apache and Letsencrypt (using Ansible - playbook could need some updates!)
add the instances to Collector
wait for a few days to spot any issues with the new deployment
if all ok, archive the logs of the old instances on metrics-store
shut the old instances down & ask the admin team to remove their DNS entries
Also just a note to say I'm going to be away to a conference for most of this week and the next, if you want to start sooner because of #40038 (closed) and have any questions I'll still be around via email.
hk-6a and de-6a should be ok for longer, the disks are sized to capture 2 years of measurements.