- Notifications
You must be signed in to change notification settings - Fork5
Commit39fc1fb
committed
Remove logic in XactLockTableWait() that attempted to mark a crashed
transaction as aborted. Since we only call XactLockTableWait on XIDsthat we believe to be currently running, the odds of this code everactually firing are minimal. It's certainly unnecessary, since atransaction that's not either running or committed will be presumedaborted anyway. What's more, it's not hard to imagine scenarios wherethis could result in corrupting pg_clog: for instance, if a bogus XIDsomehow got passed to XactLockTableWait. I think the code probablydates from the ancient era when we didn't have TransactionIdIsInProgress;back then it may have been necessary, but now I think it's a waste ofcycles and potentially dangerous. Per discussion with Qingqing Zhouand Karsten Hilbert.1 parent7d6d02b commit39fc1fb
1 file changed
+1
-15
lines changedLines changed: 1 addition & 15 deletions
Original file line number | Diff line number | Diff line change | |
---|---|---|---|
| |||
8 | 8 |
| |
9 | 9 |
| |
10 | 10 |
| |
11 |
| - | |
| 11 | + | |
12 | 12 |
| |
13 | 13 |
| |
14 | 14 |
| |
| |||
383 | 383 |
| |
384 | 384 |
| |
385 | 385 |
| |
386 |
| - | |
387 |
| - | |
388 |
| - | |
389 |
| - | |
390 |
| - | |
391 |
| - | |
392 |
| - | |
393 | 386 |
| |
394 | 387 |
| |
395 | 388 |
| |
| |||
421 | 414 |
| |
422 | 415 |
| |
423 | 416 |
| |
424 |
| - | |
425 |
| - | |
426 |
| - | |
427 |
| - | |
428 |
| - | |
429 |
| - | |
430 |
| - | |
431 | 417 |
| |
432 | 418 |
| |
433 | 419 |
| |
|
0 commit comments
Comments
(0)