-
Notifications
You must be signed in to change notification settings - Fork 4.6k
delegatingresolver: add default port to addresses #8613
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
delegatingresolver: add default port to addresses #8613
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #8613 +/- ##
==========================================
- Coverage 82.00% 81.50% -0.50%
==========================================
Files 415 415
Lines 40697 40873 +176
==========================================
- Hits 33372 33312 -60
- Misses 5937 6139 +202
- Partials 1388 1422 +34
🚀 New features to boost your workflow:
|
| addr, err := parseTarget(target.Endpoint()) | ||
| if err != nil { | ||
| return nil, fmt.Errorf("delegating_resolver: invalid target address %q: %v", target.Endpoint(), err) | ||
| } |
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 think we should guard this call to parseTarget behind the feature flag since it can potentially lead to new/different failures after this PR is merged.
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.
We can have only one block guarded by the feature flag if we instead do the following:
addr := target.Endpoint()
if target.URL.Scheme == "dns" && !targetResolutionEnabled && envconfig.AddDefaultPort {
addr, err = parseTarget(target.Endpoint(), defaultPort)
if err != nil {
// return some error
}
}This way we can use the parseTarget function from the dns resolver without any modifications.
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.
My thinking was we should do the basic check and add localhost for all addresses if no host is present irrespective of the resolver being used, but adding default port 443 is only for DNS resolver with target resolution disabled. And the first part should not be behind the flag , but only 2nd part should. That is my understanding, let me know if I am missing something.
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.
From my understanding, only the DNS resolver was defaulting the hostname to localhost. Applying this logic to all resolvers would be a behaviour change which could potentially break some custom resolver that isn't expecting this. I would recommend avoiding such a change to be safe.
Also, since we're providing a env variable flag, we should ensure the flag can revert all (if not most) changes in this PR.
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.
Changed.
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
| name string | ||
| target string | ||
| wantConnectAddress string | ||
| wantErrorSubstring string |
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.
Can you add another test param for the env variable and tests cases for the env variable being disabled?
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 already have a seperate test for it : TestDelegatingResolverEnvVarForDefaultPortDisable
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.
Can we combine the tests using an additional param? That should help reduce duplicate code.
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.
Done
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
| disableDefaultPortEnvVar bool | ||
| }{ | ||
| { | ||
| name: "no port in target", |
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.
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.
Changed.
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.
Subtest names still have spaces which make them hard to search for when you get here from a failing test run.
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.
Ohhh... I misunderstood some parts. Got it!
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 still see subtest names with spaces. The gotip that I linked to recommends that we not have spaces in them, but have underscores. That way, when you have TestFoo with a subtest name subtest_name, it will show up in the output as TestFoo/subtest_name and you can directly copy-paste and search in the code. But if you have TestFoo with a subtest name subtest name, it will show up in the output as TestFoo/subtest_name, and copy-paste will not work.
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.
Done.
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
| } | ||
| overrideTestHTTPSProxy(t, envProxyAddr) | ||
|
|
||
| targetResolver := manual.NewBuilderWithScheme("dns") |
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.
Don't you have to unregister the original dns resolver and re-register it at the end of the test, for this test to not affect other tests?
| targetResolver := manual.NewBuilderWithScheme("dns") | ||
| target := targetResolver.Scheme() + ":///" + test.target | ||
| // Set up a manual DNS resolver to control the proxy address resolution. | ||
| proxyResolver, proxyResolverBuilt := setupDNS(t) |
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.
Oh I see that this helper does re-register the original dns resolver. But maybe this section can still be a little more explicit/readable etc. I'm not able to think of the exact way to make this better right now.
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.
This is what we had earlier too..in the tests.
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
internal/resolver/delegatingresolver/delegatingresolver_ext_test.go
Outdated
Show resolved
Hide resolved
| t.Fatal("Context timed out when waiting for a state update from the delegating resolver") | ||
| // Tests the creation of a delegating resolver when a proxy is configured. It | ||
| // verifies correct error handling for invalid targets. | ||
| func (s) TestDelegatingResolverwithDNSAndProxyWithNoTargetResolutionWithErrorTargets(t *testing.T) { |
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.
Lol
I think you are going to win the longest test name award in our codebase :P :P
Maybe we can shorten these some other time.
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.
Tried to maybe shorten a bit .
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.
Tried to change it be a little shorter..Maybe?
Fixes: #8607
RELEASE NOTES:
GRPC_EXPERIMENTAL_ENABLE_DEFAULT_PORT_FOR_PROXY_TARGETfor adding a default port to addresses being sent to proxy which is set by default.