GitNimay opened a new pull request, #14053:
URL: https://github.com/apache/apisix/pull/14053
### Description
Kubernetes discovery lists Endpoints in pages of 120. `list()` in
`apisix/discovery/kubernetes/informer_factory.lua` fetches each following page
with a recursive call but ignores its result:
```lua
informer.continue = data.metadata.continue
if informer.continue and informer.continue ~= "" then
...
list(httpc, apiserver, informer)
end
return true
```
If a later page fails (for example while the API server is returning etcd
errors during a control-plane restart), `list()` still returns `true`.
`list_watch()` then runs `post_list()`, which deletes every endpoint in the
shared dict that the list did not touch, which includes everything on the pages
that never loaded. The following watch only delivers deltas, so those services
stay without endpoints (`no valid upstream node`) until APISIX restarts. No
`list failed` error is logged, which matches the report.
This PR returns the result of the recursive call. A failed later page now
fails the whole list: `list_watch()` logs `list failed`, skips `post_list()`,
and the informer retries with a full re-list. As a side effect, the recursion
is now a proper tail call.
#### Which issue(s) this PR fixes:
Fixes #14050
### Tests
Added `t/discovery/kubernetes_list_pagination.t`, which mocks `resty.http`
(same approach as `kubernetes_ssl_verify.t`) with an API server that serves the
Endpoints list in two pages:
- TEST 1: both pages succeed, so both are added and `post_list()` runs.
- TEST 2: the second page returns 500, so the list fails with `list failed`
and `post_list()` is not called.
Without the fix, TEST 2 fails (`post_list called`, `fetch_state: watch
failed`). With the fix, `kubernetes_list_pagination.t`,
`kubernetes_ssl_verify.t`, `kubernetes_cluster_ids.t` and `kubernetes_schema.t`
pass locally. `luacheck` and `lj-releng` are clean.
### Checklist
- [x] I have explained the need for this PR and the problem it solves
- [x] I have explained the changes or the new features added to this PR
- [x] I have added tests corresponding to this change
- [ ] I have updated the documentation to reflect this change (not needed,
no user-facing change)
- [x] I have verified that this change is backward compatible (If not,
please discuss on the [APISIX mailing
list](https://github.com/apache/apisix/tree/master#community) first)
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
