This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.

Bug 56005 - CPU races and memory spikes occur every 30 minutes, requiring ide restart
Summary: CPU races and memory spikes occur every 30 minutes, requiring ide restart
Status: CLOSED DUPLICATE of bug 55719
Alias: None
Product: javaee
Classification: Unclassified
Component: Code (show other bugs)
Version: 4.x
Hardware: PC Windows XP
: P1 blocker with 1 vote (vote)
Assignee: issues@javaee
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-03-04 22:27 UTC by manawiz
Modified: 2005-07-12 12:47 UTC (History)
0 users

See Also:
Issue Type: DEFECT
Exception Reporter:


Attachments
Snapshot of memory monitor showing described spikes. (2.64 KB, image/jpeg)
2005-03-04 22:28 UTC, manawiz
Details
Stack dump after memory spikes and cpu racing started (22.76 KB, text/plain)
2005-03-04 22:48 UTC, manawiz
Details
More thread/stack dumps of this problem (164.62 KB, text/plain)
2005-03-05 20:56 UTC, manawiz
Details

Note You need to log in before you can comment on or make changes to this bug.
Description manawiz 2005-03-04 22:27:17 UTC
Using NB4.1 Beta on Windows XP SP2, about every 30
minutes or so the CPU starts racing the memory
usage starts spiking up and down wildly.  There is
no recovery other than to close and restart the
IDE.  This has happened when I've done nothing in
the session other than using the edtior on jsp
files.  Another user on nbusers reports similar
symptoms on SUSE9.2 pro 64bit, so it may not be OS
specific.

I have increased the default memory settings to
-Xms64m -Xmx256m and the problem occurs long
before hitting 64m of actual usage.  I've not yet
tried increasing the default PermSize setting, but
will try that next.  I will also try to get a
stack trace on the next session and will post any
updates.

I have and will now post a jpeg of the memory
monitor that shows what I mean by "memory
spiking"; never seen anything like this in all my
time using NB.

In the last session, I was running at around 32
meg usage when all of a sudden NB took over the
CPU (99%) and the memory started shooting up and
down in rapid "spikes" as seen in the memory
monitor.  Also, the base usage started climbing
dramatically; e.g., it was less than 60 seconds
between the start of this and the snapshot you see
in the jpeg, where the base usage has increased to
 42 meg.  Spikes are increasing in size as well,
up over 80m when I stopped the ide.  The ide was
IDLE at the time this happened -- sitting in the
editor with nothing going on.

There is no cure when this occurs other than
shutting down and restarting the IDE.  I've tried
closing all the files, to no avail.

This happens on EVERY session.  I have found no
way to prevent it.  This problem renders the ide
unusable.
Comment 1 manawiz 2005-03-04 22:28:57 UTC
Created attachment 20696 [details]
Snapshot of memory monitor showing described spikes.
Comment 2 manawiz 2005-03-04 22:48:51 UTC
Created attachment 20697 [details]
Stack dump after memory spikes and cpu racing started
Comment 3 manawiz 2005-03-04 22:52:18 UTC
I've just attached a stack dump from my last session.  I increased
PermSize settings (PermSize=64m, MaxPermSize=128m).  This time the
problem occured in less than 5 minutes and I had used nothing but the
editor on one jsp file.  Don't know if the timing has anything to do
with the memory settings or not (seems unlikely), but they are
certainly no preventative.

The stack dump is attached.  I hit control-Break twice to get two
snapshots; both are in the file.
Comment 4 manawiz 2005-03-05 06:12:24 UTC
It seems that with PermSize=64m, the MTBF is consistently around 5
minutes or less.  The best MTBF is with PermSize=32m (default), but
with the Xms and Xmx cranked up as indicated earlier (64m and 256m,
respectively).  Here I can get 30m-60m of work done before NB crashes
-- still pretty poor when I've be using 4.0 for days at a time. 
Something is seriously wrong in 4.1 beta.  Also, I've tried deleting
var/cache, but this has no impact.  The problem continues...
Comment 5 manawiz 2005-03-05 20:56:47 UTC
Created attachment 20705 [details]
More thread/stack dumps of this problem
Comment 6 manawiz 2005-03-05 20:59:08 UTC
Per a request from Martin Matula on nbusers, I have attached about 10
more thread dumps made in succession (with a couple seconds between
each) from my last session when the problem arose.

Note there are a couple interesting messages above the thread dumps of
the form "[folding] endless loop detection: skipped...".  I don't know
when these messages occurred relative to the problem as the Command
window was not visible.
Comment 7 Martin Matula 2005-03-05 21:22:47 UTC
Seems that the processor spends most of the time in
JspSyntaxSupport.getScriptingChain(). Reassigning to web module for
evaluation.
Comment 8 manawiz 2005-03-05 21:27:26 UTC
If the bug is specific to jsp editing, please see issue 56024 which is
possibly related.  I file that issue against the editor module --
perhaps it belongs with the web module also?
Comment 9 Pavel Buzek 2005-03-06 16:08:11 UTC
Looks like a duplicate. The stack traces and symptoms are similar.
Please test in daily build and reopen if you still see the problem.

*** This issue has been marked as a duplicate of 55719 ***
Comment 10 manawiz 2005-03-09 19:54:23 UTC
This issue seems to have gone away in the March 6 daily build.  Thanks!  This is
a sufficienlty critical issue that users would appreciate a beta update I'm sure...