Title: Codec encodings can exhaust permgen
Type: crash Severity: major
Components: Core Versions: Jython 2.7
Status: open Resolution:
Dependencies: Superseder:
Assigned To: Nosy List: jeff.allen, zziad
Priority: low Keywords: thread-interpreter context

Created on 2016-03-13.14:22:37 by zziad, last changed 2018-03-13.22:42:55 by jeff.allen.

File name Uploaded Description Edit Remove
jythonOOM.png zziad, 2016-03-13.14:22:36 Heap dump + work around
msg10818 (view) Author: zaid (zziad) Date: 2016-03-13.14:22:35
Permgen out of memory error is thrown after performing a stress test, the root cause of this issue is loading many classes for encodings (for each interpreter) so I think we can make it shared encoding lib for all interpreters if we do not want a different encoding for each, please check the attached image which describes a work around that I made to solve this issue, it works till now without any issue.
msg11794 (view) Author: Jeff Allen (jeff.allen) Date: 2018-03-13.22:42:55
The codec state has to be provided for each interpreter. However, it is reasonable to ask whether we could not share more of the immutable data between threads when many interpreters run concurrently.
Date User Action Args
2018-03-13 22:42:55jeff.allensetkeywords: + thread-interpreter context
priority: low
messages: + msg11794
nosy: + jeff.allen
2016-03-14 23:51:06zyasoftsettitle: High memeory consumption -> Codec encodings can exhaust permgen
2016-03-13 14:22:37zziadcreate