-
Type: New Feature
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: 3.4.13, 3.6.2
-
Component/s: Diagnostics
-
Fully Compatible
-
Platforms 2018-06-18, Platforms 2018-07-02, Platforms 2018-07-16, Dev Tools 2019-09-09, Dev Tools 2019-10-21, Dev Tools 2019-11-18, Dev Tools 2019-12-02, Dev Tools 2019-12-16, Dev Tools 2019-12-30, Dev Tools 2020-01-13, Dev Tools 2020-01-27
-
(copied to CRM)
Occasionally we need to capture the state of all threads for performance analysis.
Currently this requires installing `gdb` and running an external script.
I propose implementing a signal handler calling printStackTrace().
This would ease diagnostics as we can achieve the same result by issuing a kill signal.
- depends on
-
SERVER-36178 Evaluate adding -fasynchronous-unwind-tables to our build flags
- Closed
-
SERVER-36242 Use libunwind for backtraces
- Closed
- is depended on by
-
SERVER-34190 MongoDB process hangs after some random time
- Closed
- is related to
-
SERVER-48396 SIGUSR2 stack collection closes connections
- Closed
-
SERVER-33880 Log currentOp information to mongod log
- Backlog
-
SERVER-31814 Log details of process which sent signal (in Linux)
- Closed