Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Update approach to how we manage and configure the weave network #395

Closed
tayzlor opened this issue Jul 16, 2015 · 8 comments
Closed

Update approach to how we manage and configure the weave network #395

tayzlor opened this issue Jul 16, 2015 · 8 comments
Assignees
Milestone

Comments

@tayzlor
Copy link
Member

tayzlor commented Jul 16, 2015

See http://blog.weave.works/2015/07/16/bridge-over-troubled-weavers/

As part of this we also need to update the docs here https://github.com/Capgemini/Apollo/tree/master/roles/weave

and the blog post here http://capgemini.github.io/devops/how-apollo-uses-weave

@enxebre
Copy link
Contributor

enxebre commented Jul 23, 2015

Dropped a comment here
weaveworks/weave#117 (comment)

@enxebre
Copy link
Contributor

enxebre commented Jul 23, 2015

this #325 needs to be fixed as part of this ticket.

@pmbauer
Copy link
Contributor

pmbauer commented Sep 10, 2015

At udacity, we've observed an extremely high price for using the weave overlay network for container-to-container communication. (aws, vpc).
Using qperf, we've measured the following relative bandwidths (1x is over VPC)
eth0-hostA -> eth0-hostB : 1x
eth0-hostA -> weave-hostB: .5x
weave-hostA -> weave-hostB: .05x

Dial-up modem speeds over the weave overlay network is unusable.

So we will be spending time in our following sprint to sort this out. Just checking if anyone else had observed this and had tried out weave-managed device and weave-allocated IP blocks (per the troubled-weavers post); didn't want to duplicate work.

@enxebre
Copy link
Contributor

enxebre commented Sep 10, 2015

Hi, we've been waiting for weaveworks/weave#1376 to get merged as otherwise the new weave approach (not overriding docker dockerbridge) will break the consul/registrator/weave discovery ecosystem (see weaveworks/weave#117 (comment))
The "fix" is included in new version of weave so this new approach could be tested now.

We also want to move to fast datapath http://blog.weave.works/2015/06/12/weave-fast-datapath/ which is unstable and linked to an old version of weave at the moment but hopefully will have some improvements in the next release of weave.
To answer your question there is no work in progress for this at the minute on this side.

@pmbauer
Copy link
Contributor

pmbauer commented Sep 10, 2015

@enxebre Thanks for the summary; I'll report back when we try the new version of weave, probably late next week.

@enxebre enxebre removed the blocked label Oct 22, 2015
@enxebre
Copy link
Contributor

enxebre commented Oct 22, 2015

Hey @pmbauer we've just made a priority of this as well as moving to fastdatapath which It's just been released https://github.com/weaveworks/weave/releases/tag/v1.2.0
So we are starting to work on this, any feedback from your side would be appreciate :)

@pmbauer
Copy link
Contributor

pmbauer commented Oct 22, 2015

@enxebre nice! things happen at work as they do. A bunch of other tasks trumped our weave re-work so I don't have anything to report yet unfortunately.

@enxebre
Copy link
Contributor

enxebre commented Nov 4, 2015

closing as this got it #504

@enxebre enxebre closed this as completed Nov 4, 2015
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants