Repository navigation
Commit fb8db61
[fix](test) wait for delete visibility instead of fixed sleep in test_f_delete_publish_skip_read
The fixed Thread.sleep(12000) before the normal read is racy. When the disable_block HTTP request is queued behind the in-flight flexible stream load on the same BE accept thread, the publish spin only clears at the FE publish timeout (~152s), so the delete (version 4) becomes visible after the read and k1=2 stays visible (3 rows instead of 2). Replace the sleep with a bounded Awaitility poll until the delete is actually published and visible (k1=2 gone). The normal-read count is monotonic 3 -> 2, so this is a stable signal and fails loudly on timeout if the delete were genuinely not applied. Not a product bug; the storage engine applied the delete correctly.
Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>1 parent fbfd8fb commit fb8db61
1 file changed
Lines changed: 7 additions & 1 deletion
File tree
- regression-test/suites/fault_injection_p0/flexible/legacy
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
| |||
101 | 101 | | |
102 | 102 | | |
103 | 103 | | |
104 | | - | |
| 104 | + | |
| 105 | + | |
| 106 | + | |
| 107 | + | |
| 108 | + | |
| 109 | + | |
| 110 | + | |
105 | 111 | | |
106 | 112 | | |
107 | 113 | | |
| |||
0 commit comments