Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • Sep 07 08:46
    designermonkey commented #641
  • Sep 05 16:16
    matija commented #643
  • Sep 04 10:37
    nohaapav commented #644
  • Sep 04 10:34
    nohaapav commented #641
  • Aug 25 02:39
    osu commented #644
  • Aug 25 02:38
    osu opened #644
  • Aug 24 22:00
    DannyDevil0 commented #634
  • Aug 13 07:10
    louislam commented #643
  • Aug 11 15:54
    designermonkey commented #641
  • Jul 21 09:17
    NgHuuAn commented #560
  • Jul 21 09:17
    NgHuuAn commented #560
  • Jun 24 08:58
    i4j5 opened #643
  • Jun 23 11:42
    i4j5 commented #642
  • Jun 21 16:06
    lumir-mrkva commented #642
  • Jun 21 02:17
    anhhtca commented #635
  • Jun 20 16:52
    i4j5 commented #323
  • Jun 20 16:47
    i4j5 commented #323
  • Jun 20 16:45
    i4j5 commented #635
  • Jun 20 13:38
    i4j5 opened #642
  • Jun 04 08:37
    designermonkey commented #641
Chase Pierce
@syntaqx
I'm trying to figure out what I need to configure for my services to show up correctly
Chase Pierce
@syntaqx
@nohaapav Any update on the terraform provider/api docs? :)
Pavol Noha
@nohaapav
not really ... busy with other work :) We're not using datadog in current setup ..
Chase Pierce
@syntaqx
FeelsBad. Hoping you guys will release stuff soon :(
Any way I can help? :)
Pavol Noha
@nohaapav
Nothing on my mind apart of 1.7 release .. But again clojure knowledge is a must .. :)
Chase Pierce
@syntaqx
I'm slowly learning it ;)
Chase Pierce
@syntaqx
@nohaapav How goes development of the various things? Terraform provider, api docs, general happiness I can start using? :)
Pavol Noha
@nohaapav
Finalizing release .. not really time for terraform .. api doc will be probably done in 1.8 (we're going to integrate swagger but first we need to introduce reitit instead of bidi routing)
Chase Pierce
@syntaqx
I suppose that's fair, hah
Ghost
@ghost~530e888d5e986b0712efbd49
I need some help with understanding how the auto redeploy works.
I cant seem to find any documentation for swarmpit
Pavol Noha
@nohaapav
If new image is pushed in registry, swarmpit update the service with force flag accordingly -> new image is used instead of old one ..
Ghost
@ghost~530e888d5e986b0712efbd49
how is it detecting which version to update to?
i have app:dev-1.0.0
does that mean if I push app:dev-1.0.1 it will use that next?
Chase Pierce
@syntaqx
@bludot It watches for when a tag gets overwritten, ie, image:latest. When something overwrites that, it redeploys it - Swarmpit doesn't leverage versions, so you also lose rollbacks by using this
Ghost
@ghost~530e888d5e986b0712efbd49
ah ok ok. so always use latest
Chase Pierce
@syntaqx
Just whatever tag you deploy is the one it's going to watch
Pavol Noha
@nohaapav
exactly
Ghost
@ghost~530e888d5e986b0712efbd49
thank you
Chase Pierce
@syntaqx
How goes things in delivery world? Are we getting any closer to having some of that new hotness terraform and such?
Pavol Noha
@nohaapav
Hi Chase, not really .. we're working on time series monitoring and alerts, dashboards & autoscalling ..
Chase Pierce
@syntaqx
:cry:
rwestbrookjr
@rwestbrookjr
Is anyone online?
godavarthiashok
@godavarthiashok
Hi Everyone
Regarding swarmpit, how does load balancing works in Swarmpit ,
Should we mention each port for each replica in docker compose yml or mentioning a single port will also help , please suggest.
Pavol Noha
@nohaapav
@godavarthiashok Hi, sorry .. we had vacation recently and our sync agent with gitter went down ..
You should always have just one replica of swarmpit ..
Swarmpit is management tool so short outage is ok .. Swarm is going to redeploy new instance almost immediately because of default restart policy ..
@rwestbrookjr hi finally yes :)
Mike
@mhsutton_gitlab
I just moved from portainer to swarmpit - wow, just incredibly better for me - faster, feels more robust and responsive.
I've been through a baptism of fire with containerisation - happy to help anyone here with traefik, gluster, loads of stuff
Mike
@mhsutton_gitlab
@nohaapav - is there any info on how the redeploy when image updated feature works, can it be configured etc. There are some images I want to control when they are updated in my swarm (currently I use sheperd).
@nohaapav is it 'AutoRedeploy' on the service definition?
lumir-mrkva
@lumir-mrkva
@mhsutton_gitlab swarmpit check regularly all services that have swarmpit.service.deployment.autoredeploy label set and compares their deployed tag hash with hash in the repository and triggers update if it differs
that way your CI doesn't have to have access to your infrastructure to deploy new version
btw. we're planning to have some recommended example compose files of traefic and stuff to get you started in repo, did you setup traefik 2.0 ?
Matthias Baldi
@matthiasbaldi
I'll try it soon to migrate traefik to 2.0. I run into the "auto-migration" 😬😝
lumir-mrkva
@lumir-mrkva
@matthiasbaldi i am currently setupping new cluster utilizing traefik 2 so i should be able to share some stackfiles soon, when i get it working properly
Matthias Baldi
@matthiasbaldi
👍 👌
Mike
@mhsutton_gitlab
@lumir-mrkva - haven't made the move to traefik 2.0 yet, scheduled to look into that in our lab cluster in 2-3 weeks. Things just got stable
also - I see the 'View Logs' on services, is there a way to run a command - like bash to get into a running container?
Pavol Noha
@nohaapav
@mhsutton_gitlab Hi Mike, you can't exec into container via Swarmpit at the moment.
Mike
@mhsutton_gitlab

perhaps a little off topic but... one postgres per app database or one shared postgres container with multiple app databases?

shared container pros: easier to manage, backup, replicate. Cons: single point of failure (all apps down if pg container goes down)
per app postgres pros - resilience/app isolation; Cons: different versions, more stuff to maintain, replicate

How do you decide?

Mike Holloway
@mh720
@mhsutton_gitlab scaling is possibly the concern for pointing all your apps at the same db instance. If the load from your multiple apps is easily served by a single postgres instance, this might be desired due to the management pros you shared. As your load grows over time, you could always spin up another db instance and migrate data for one of your busy DBs into it and re-point apps.
Mike
@mhsutton_gitlab
@mh720 yeah - really good point, unfortunately instance scaling is not an option yet (because persisted volumes are too slow) and actually having more replica nodes makes no difference to capacity. I think what you outline is the only way to share the load. Which is not a bad option, things will have to have gotten seriously busy though. The replica stuff already provides replication on servers in a different geo-zone
@mh720 thanks for the response
rajaws2015
@rajaws2015
Hi Guys, i am new to Swarmpit. Currently we have running swarmpit for my production based docker swarm.