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
Make local VM host IPs '10.0.2.2' configurable #1468
Conversation
Codecov Report
@@ Coverage Diff @@
## master #1468 +/- ##
=======================================
Coverage 56.13% 56.13%
=======================================
Files 54 54
Lines 6362 6362
=======================================
Hits 3571 3571
Misses 2791 2791 Continue to review full report at Codecov.
|
Consider this highlight in lieu of requesting a review from @AdamWill which is conceputally impossible. |
Sorry, I can't parse your english :) Could you rephrase for me please? |
I can't request a review from Adam for this pull request. So I used a highlight. |
Hum. So let's see...
Smaller notes:
|
Well, the os-autoinst-openvswitch part is probably independant of the qemu usermode networking bits so I think it's better if I leave these changes out for now. About setting according qemu parameters, sure we could do this but I thought we could try one thing at a time. After all I did not really test this and would rely on you ;) |
"Well, the os-autoinst-openvswitch part is probably independant of the qemu usermode networking bits" Hmm, I'm kinda struggling with this. You can kinda see it both ways. But I'm not sure how you can really use this as-is, at least if you have any tap tests, because - where are you going to set |
yes, you are right. I haven't considered this. I misinterpreted the line |
Well, I don't think it's impossible. I think your initial proposal would work if we deal with the rewrite issue (perhaps by adopting my version which only allows for one specific alternate configuration?) and documented it properly ("you have to set I think any solution here is gonna have some slightly rough edges, but we're working towards something that's manageable as long as we document it properly, and make sure the simple case (just leave everything at 10.0.2.2) remains simple. WDYT? |
oh, I just remembered, in your proposal |
I agree with both your latest proposal. In general I would also foresee environment variables + test variables together with a clear precedence (test > env > default). So, where can we see your version then? :) |
I should've seen that one coming :P I'll try and work something up combining your basic framework with my half-done version, with BONUS ADDED DOCS, this week. |
I would not fine with just very rough, hacky, dirty, incomplete code snippets |
AdamWill@010fdb4 is what I had so far. I haven't tested it yet, but that's where I was going. |
d0904d1
to
f78218c
Compare
Ok, so what I have done now is simply make the "10.1.0.0" rewrite target configurable by environment variable as well and mentioned that in doc/networking.md . I would prefer to be not too smart in the scripts and not try to guess the rewrite target based on ip and netmask but rather ask for explicit configuration. Would this work for you? |
It should be workable, yeah. If we're going to make the admin do all the work it'll be good to document it carefully I guess. I'll try this out today and see how it goes - I still didn't actually test any version of this yet so maybe there are bear traps we haven't spotted yet :P |
So I tried this out briefly today but it didn't go entirely smoothly. It did seem like two tests that took 172.16.2.x IPs could talk to each other, but tests didn't seem to be able to reach the public internet or, perhaps, the host IP. May well be errors on my part in the local config rather than anything in the PR, but thought I'd mention it. It's very Friday right now, so I'm gonna try it again on Monday. |
Ok, good to know. I am awaiting further test results of yours |
So I found at least one problem in this: there's a sneaky hidden hardcoded IP address we missed. In the ARP rule, So, here's my fix for that:
...
I'm testing this out now. |
OK, I have my modified version deployed to Fedora production and it appears to be working. Yay. |
cool, updated my commit accordingly. |
Thanks, I think you missed a bit though - it needs to be |
Yeah, sorry. Got that wrong. Should have used a verbatim copy :D |
Related progress issue: https://progress.opensuse.org/issues/68851
ready |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I suppose it is good to merge. (I haven't followed the entire conversation and all the details of the progress issue.)
yeah it was good to go. our instance is still working fine with this. |
Related progress issue: https://progress.opensuse.org/issues/68851