| Bug #77597 | LocalDLFifoList not destroyed properly | ||
|---|---|---|---|
| Submitted: | 2 Jul 2015 10:45 | Modified: | 14 Jan 2016 22:40 |
| Reporter: | Mikael Ronström | Email Updates: | |
| Status: | Closed | Impact on me: | |
| Category: | MySQL Cluster: Cluster (NDB) storage engine | Severity: | S2 (Serious) |
| Version: | 7.4.7 | OS: | Any |
| Assigned to: | CPU Architecture: | Any | |
[14 Jan 2016 22:40]
Jon Stephens
Fixed in NDB 7.4.8. See BUG#77612 for documentation info. Closed.
Description: When a scan is closed and another started one uses a LocalDLFifoList to search for all queued scans. The problem with the LocalDLFifoList is that it maintains the queue state on the stack, so if one calls a function before destroying this object then havoc occurs. This happens when running test case developed for BUG#20981491. How to repeat: Run test case developed for BUG#20981491 Suggested fix: Add another {}-pair such that LocalDLFifoList is destroyed at a time before calling function that can manipulate the base object used to create LocalDLFifoList.