Commit a8c5ae0
authored
Don't try to use a fixed port in test_server_config_default (#8272)
We had a flake regarding this.
(At least on my machine?), linux never randomly assigns an even
numbered port, so probably the flake collided with something external
specifically looking for it?
The find_available_port mechanism is fundamentally racy, of course;
the port could be allocated by someone else immediately, and in
testing I did manage to observe cycle lengths as short as 2 before a
reuse.
Possibly these tests should just have a big hammer
`@retry_failed_test(3)` decorator around them or something, but I am
reluctant to introduce this because I fear we will use it instead of
writing proper tests.
(Most tests don't have as good an excuse for flakiness as needing to
allocate a shared resource in a 16-bit namespace).1 parent 0676b5e commit a8c5ae0
2 files changed
+6
-19
lines changed| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
3002 | 3002 | | |
3003 | 3003 | | |
3004 | 3004 | | |
3005 | | - | |
3006 | | - | |
3007 | | - | |
3008 | | - | |
3009 | | - | |
3010 | | - | |
3011 | | - | |
3012 | | - | |
3013 | | - | |
3014 | | - | |
3015 | | - | |
3016 | | - | |
3017 | | - | |
3018 | | - | |
3019 | | - | |
3020 | | - | |
3021 | | - | |
| 3005 | + | |
| 3006 | + | |
| 3007 | + | |
| 3008 | + | |
3022 | 3009 | | |
3023 | 3010 | | |
3024 | 3011 | | |
| |||
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
2366 | 2366 | | |
2367 | 2367 | | |
2368 | 2368 | | |
2369 | | - | |
| 2369 | + | |
2370 | 2370 | | |
2371 | 2371 | | |
2372 | 2372 | | |
| |||
2378 | 2378 | | |
2379 | 2379 | | |
2380 | 2380 | | |
2381 | | - | |
| 2381 | + | |
2382 | 2382 | | |
2383 | 2383 | | |
2384 | 2384 | | |
| |||
0 commit comments