Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Mar 25 05:30
    a-parser opened #18
  • Nov 21 2020 18:00

    mh720 on master

    karma v0.74, kthxbye v0.9 (compare)

  • Sep 14 2020 17:50

    mh720 on master

    karma to 0.70, cadvisor to 0.36 (compare)

  • Sep 03 2020 11:16
    saulmmendoza commented #17
  • Sep 03 2020 11:10
    saulmmendoza commented #14
  • Jul 06 2020 14:14

    mh720 on master

    kthxbye v0.68 (compare)

  • Jun 23 2020 19:26

    mh720 on master

    karma 0.60 -> 0.65, kthxbye 0.6… (compare)

  • Apr 10 2020 16:36

    mh720 on master

    updated portworx volume status … rev karma to v0.60 (compare)

  • Mar 04 2020 21:19

    mh720 on master

    cadvisor 0.35, Portworx 2.4.0 (compare)

  • Mar 04 2020 20:28

    mh720 on master

    Portworx px-dev 2.4.0 (compare)

  • Feb 21 2020 18:29

    mh720 on master

    move from caddy:no-stats-1.0.3 … (compare)

  • Feb 20 2020 20:11

    mh720 on master

    rev kthxbye to 0.6, add max-dur… (compare)

  • Jan 30 2020 16:25

    mh720 on master

    karma 0.54 > 0.55 (compare)

  • Jan 29 2020 19:34

    mh720 on 1.9.1

    (compare)

  • Jan 29 2020 19:06

    mh720 on master

    add up script to update service… (compare)

  • Jan 28 2020 18:34

    mh720 on master

    recommend VictoriaMetrics inste… (compare)

  • Jan 27 2020 22:22

    mh720 on master

    update netdata-based Nodes and … (compare)

  • Dec 12 2019 18:05

    mh720 on master

    cadvisor v0.34.0, karma v0.52 cadvisor v0.34.0, karma v0.52 (compare)

  • Dec 02 2019 23:31
    mh720 commented #17
  • Dec 02 2019 23:29
    mh720 commented #17
Mike Holloway
@mh720
Ask questions or discuss feature requests here
Mike Holloway
@mh720
Just validated that swarmstack and errbot deploy well on Google Compute VM instances , without DNS and just using their external IP as the EXTERNAL_ENDPOINT. This is good!
Mike Holloway
@mh720
Mike Holloway
@mh720
Just added a docker-compose-singlebox.yml stack that can be used to test out the swarmstack DevOps tools stack without installing Portworx or other dedicated storage. Deploys on a mac in just minutes.
JM Lopez
@jm66
Hi @mh720, I wonder why did you replace node-exporter with netdata?
Mike Holloway
@mh720
I was just thinking about that as a matter of fact, and was considering mentioning ADDING it in the scaling section. NetData suited the system resources data collection with low CPU usage (<1% cpu core) and super-high resolution compared to node_exporter, and supports a fairly robust plugin system of it’s own. NetData also provides out-pf-the-box metrics visuals as well. Newer users to the Prom ecosystem may find utility in NetData’s UI, scaling users may later remove NetData, replace Grafana Dashboards, default rules, etc. Node-exporter can be added where needed, and other pieces of swarmstack can be commented out and replaced with scaled versions of those components at a later time if the deployment requires it. The cpu footprint and resolution that NetData provided was the initial reason I went with it over node-exporter.
Mike Holloway
@mh720
And node-exporter is limited when run in a Docker container, Docker stacks currently don’t provide some of the privileged container capabilities (moby/moby#32801) that may be needed to read cgroups, etc on nodes. That was what got me looking at NetData in the first place.
You can go through the work and get node-exporter working outside of stacks or on the node, but NetData provides decoupled and easy installation on various OSes.
Mike Holloway
@mh720
Mike Holloway
@mh720
swarmstack 0.9.8 has just been released, and replaces unsee with karma, adds Docker service healthchecks for many of the containers, added a SCALING section in the README.md, and added an InfluxDB compose file with documentation for both swarmstack and non-swarmstack users. YouTube vids
Mike Holloway
@mh720
3 Grafana dashboards received about 15 hours of work in 0.9.8-2 release
Mike Holloway
@mh720
0.9.9 out now, with examples of prometheus scraping for Caddy proxys. Deploy stacks behind their own tiny Caddy instances, and have Prometheus scrape them automatically by adding prometheus.monitoring labels to your own application Caddy services.
fward1208
@fward1208
Is it possible to tie PMM into swarmstack and use the current prometheus configuration, or would you need to use the one that PMM natively installs?
Mike Holloway
@mh720
Not really sure what they are doing inside their docker image or virtual appliance, but I don’t see why the swarmstack prometheus couldn’t be a datasource for their Grafana. Or have their Prometheus scrape the similar targets that swarmstack would configures it’s own Prom for.
fward1208
@fward1208
Ok, thanks for the reply, Mike. I know they try to force a new Prometheus install. I've been trying to find any information about the possibility and apparently it is a subject that has not really been explored to date.
Mike Holloway
@mh720
In that case, you could just comment out the prometheus service stuff in swarmstack/docker-compose.yml and swarmstack will generate the prometheus/conf/prometheus.yaml for you of all the scrape targets, and maybe you can just put those into their Prom.
fward1208
@fward1208
I will give that a shot. Thanks again, Mike.
Mike Holloway
@mh720
you bet!
fward1208
@fward1208
Hey, Mike are you around? I was wondering if it might be possible to convert the iptables rules to firewall-cmd instead. I noticed the rules files state that any edits will be replaced by ansible.
Mike Holloway
@mh720
I welcome any PRs towards giving swarmstack users options for choosing alternate tools that do the same job. I’m only personally familar with programming the kernel firewall chains around the intricacies of the iptables user-land tools.
You can ignore the ansible warning if you don’t use the firewall.yml playbook.
Each swarmstack playbook is optional and should be useful on it’s own merits, such as just for firewall management of nodes, installing/updating docker, etc.
fward1208
@fward1208
Ok, thanks, Mike.
Chase Pierce
@syntaqx
Anyone have any advice on connecting CircleCI to a swarm cluster so it can perform a rolling update? That, or how do you guys CI your updates?
Mike Holloway
@mh720
Have a look at Portainer, which offers webhooks to restart/freshen containers/stacks
@syntaqx
Mike Holloway
@mh720
Can also apply an update delay (3m, 1h, etc) to swarm services to bring changes in a rolling-update manner: https://docs.docker.com/engine/swarm/swarm-tutorial/rolling-update/
Chase Pierce
@syntaqx
Seems like the portainer stuff just auto-updates a given tag, no? (ie, my/service:latest)
Isn't the more common convention in orchestration to use a specific tag (ie my/service:1.0.1) so you can have rollbacks?
Mike Holloway
@mh720
For dev CI/CD we use latest and fail forward, for production a tagged release. Not a fit for every environment, but is adequate for some.
Mike Holloway
@mh720
Curious to hear others suggestions in gitter/docker/swarm
Chase Pierce
@syntaqx
Anyone using datadog agent to collect logs in their swarm?
Trying to figure out what labels I need to add to services - the agent appears to be running.
Mike Holloway
@mh720
https://github.com/swarmstack/loki documents container log collection for Docker swarms directly into Grafana/Loki
Also announcing https://github.com/swarmstack/teampass which provides a working Docker compose file for the TeamPass collaborative password manager.
Chase Pierce
@syntaqx
Woo fancy
Mike Holloway
@mh720
Lol, yep, useful for recording shared credentials between team members in a secure and revokable way. Now this is fancy: currently testing https://github.com/Comcast/trickster to cache prometheus queries for (Grafana) dashboards; it reduces prometheus load effectively when larger teams have many watchers of a dashboard with 100+ metrics being re-queried every update period, per-person. Will likely release a compose file and documentation for swarmstack users soon.
Mike Holloway
@mh720
https://github.com/swarmstack/trickster caching proxy is now available to swarmstack users, easily reduce load on Prometheus from popular / complex dashboards
Mike Holloway
@mh720
https://github.com/swarmstack/influxdb a contributor just released a Grafana dashboard for visualizing pretty much all InfluxDB OSS ‘internal’ metrics
matrixbot
@matrixbot
lub can I somehow attach my host to a overlay network to address containers directly from the host?
Mike Holloway
@mh720
Overlay networks (and service dns names) are only available from inside containers. To connect to a service from a host, you’d need to expose the service port, then your host(s) can connect to it by addressing the host’s own address and that port. Any host IP address in the swarm can be the target address and will forward they traffic from that exposed port into the service container(s).
François Leurent
@131
Hi all