You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to run run_custom against treeline under two threads. It yields a segmentation fault. After debugging with gdb, I find there is suspected thread synchronization error.
Steps to reproduce:
Build the project with debug mode on
Run gdb ./run_custom
Run with parameters --workload_config=../../bench/workload_configs/phased_64B_A_B_A_B_A.yml --threads=2 --db="treeline"
A segmentation fault is triggered. A rough examination of the corresponding code suggests that the address of the variable pool_ is problematic. Further debugging with gdb indicates that it functions well in one thread while it is assigned invalid address in the other thread.
The issue #74 and the pull request #76 report and address a similar problem respectively. However, the fix is for pg-treeline only. I tried to run pg-treeline under the same circumstances and no errors are reported, while this is not the case for treeline.
The text was updated successfully, but these errors were encountered:
Hey @levitar64 - thanks for giving TreeLine a try and for looking into this error!
pg_treeline is actually the primary TreeLine artifact. When you run using --db=treeline, you'll run against an earlier version of the system that we're not actively supporting. We are working on refactoring the code to rename the systems to avoid this mixup (#102, #103).
Long story short - please stick to the pg_treeline config until we get around to retiring the --db=treeline config.
I am trying to run
run_custom
against treeline under two threads. It yields a segmentation fault. After debugging with gdb, I find there is suspected thread synchronization error.Steps to reproduce:
gdb ./run_custom
--workload_config=../../bench/workload_configs/phased_64B_A_B_A_B_A.yml --threads=2 --db="treeline"
pool_
is problematic. Further debugging with gdb indicates that it functions well in one thread while it is assigned invalid address in the other thread.The issue #74 and the pull request #76 report and address a similar problem respectively. However, the fix is for
pg-treeline
only. I tried to runpg-treeline
under the same circumstances and no errors are reported, while this is not the case fortreeline
.The text was updated successfully, but these errors were encountered: