Task #12049 (closed)
Opened 10 years ago
Closed 10 years ago
Prevent excessive retention of BufferedReaders by ServiceHandler
| Reported by: | jballanco-x | Owned by: | jballanco-x |
|---|---|---|---|
| Priority: | critical | Milestone: | 5.0.1 |
| Component: | General | Version: | 5.0.0 |
| Keywords: | search, full text indexing | Cc: | |
| Resources: | n.a. | Referenced By: | n.a. |
| References: | n.a. | Remaining Time: | n.a. |
| Sprint: | n.a. |
Description
The fix for #11936 involved aggressively closing the java.io.BufferedReader? instances used to read parsed files as soon as parsing is finished. However, because of the way that cleanup messages are handled by the ServiceHandler?, references to each of these objects is still retained for the duration of batch processing by the ServiceHandler?. Over time, even though these objects are closed they are putting enough pressure on the heap that they lead to OOMs.
We should prevent the ServiceHandler? from retaining these references when they are already handled by the ParserSession?.
Change History (2)
comment:1 Changed 10 years ago by Josh Moore <josh@…>
comment:2 Changed 10 years ago by jballanco-x
- Resolution set to fixed
- Status changed from new to closed
BufferedReader instances are being cleaned up properly, and search index is still functioning. Not completely clear if this will finally resolve heap memory issues, but for the scope of this ticket the work is complete. Closing.
(In [38ae08d98d3bf532ef3bf83a1efed7e874b6d840/ome.git] on branch develop) Merge pull request #2156 from jballanc/buffered-reader-oom-fix
Prevent ServiceHandler from retaining excess java.io.BufferedReader instances (see #12049)