Ticket #5466 (closed: fixed)
Probable memory leak(s) highlighted by live data run
| Reported by: | Russell Taylor | Owned by: | Russell Taylor |
|---|---|---|---|
| Priority: | major | Milestone: | Release 2.2 |
| Component: | Mantid | Keywords: | |
| Cc: | Blocked By: | ||
| Blocking: | Tester: | Wenduo Zhou |
Description
After running a live data session over a weekend, memory usage was ~25GB even though this was a simple example (and was not preserving events). Check what's going on and fix it.
The run used the FakeEventDataListener to generate 100k events/sec. The class was modified to load in the CNCS instrument. The pre-processing step was a rebin (with PreserveEvents on) and there was no post-processing step. The accumulation method was an add.
Change History
comment:2 Changed 8 years ago by Russell Taylor
Re #5466. Clear a couple of memory leaks.
These were highlighted by running the LoadLiveData unit test. They won't affect the situation that led to the creation of this ticket as they're MD-related and my test run wasn't.
Changeset: d6cbca26ff06666e9326f222e3dca3ba1265a6d6
comment:3 Changed 8 years ago by Russell Taylor
Re #5466. A member variable wasn't being initialised.
This was leading to a likely memory leak in the EventWorkspace MRU list.
Changeset: 85a7225500b0d7d92801486b9683c9d251204c89
comment:4 Changed 8 years ago by Russell Taylor
Re #5466. Remove member variable that isn't used, but is leaked.
Changeset: a5de66e1e948cdb5498119eeed59e0afad008e36
comment:5 Changed 8 years ago by Russell Taylor
Re #5466. Clear a couple of memory leaks.
These were highlighted by running the LoadLiveData unit test. They won't affect the situation that led to the creation of this ticket as they're MD-related and my test run wasn't.
Changeset: d6cbca26ff06666e9326f222e3dca3ba1265a6d6
comment:6 Changed 8 years ago by Russell Taylor
Re #5466. A member variable wasn't being initialised.
This was leading to a likely memory leak in the EventWorkspace MRU list.
Changeset: 85a7225500b0d7d92801486b9683c9d251204c89
comment:7 Changed 8 years ago by Russell Taylor
Re #5466. Remove member variable that isn't used, but is leaked.
Changeset: a5de66e1e948cdb5498119eeed59e0afad008e36
comment:8 Changed 8 years ago by Russell Taylor
Things look much better now. Running overnight ended up with a memory usage of ~500MB, which is much better than the previous situation where you could immediately see the memory usage steadily climbing. The change in [85a72255] was the one that fixed things.
There was still a growth of about 200MB in memory usage, suggesting some small leaks may still be present but nothing serious.
comment:9 Changed 8 years ago by Russell Taylor
- Status changed from accepted to verify
- Resolution set to fixed
comment:10 Changed 8 years ago by Russell Taylor
Re #5466. Clear a couple of memory leaks.
These were highlighted by running the LoadLiveData unit test. They won't affect the situation that led to the creation of this ticket as they're MD-related and my test run wasn't.
Changeset: d6cbca26ff06666e9326f222e3dca3ba1265a6d6
comment:11 Changed 8 years ago by Russell Taylor
Re #5466. A member variable wasn't being initialised.
This was leading to a likely memory leak in the EventWorkspace MRU list.
Changeset: 85a7225500b0d7d92801486b9683c9d251204c89
comment:12 Changed 8 years ago by Russell Taylor
Re #5466. Remove member variable that isn't used, but is leaked.
Changeset: a5de66e1e948cdb5498119eeed59e0afad008e36
comment:13 Changed 8 years ago by Wenduo Zhou
- Status changed from verify to verifying
- Tester set to Wenduo Zhou
comment:15 Changed 5 years ago by Stuart Campbell
This ticket has been transferred to github issue 6312
