Some component is leaking memory. I have been suspecting that for some
time. What we need is for somebody to run some heap profiles on long
running nodes.
See: http://www.javaworld.com/javaworld/jw-12-2001/jw-1207-hprof.html
On Thu, Sep 05, 2002 at 04:05:42PM -0400, Dan Merillat wrote:
>
>
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/cbe7bd07/attachment.pgp>
On Thu, Sep 05, 2002 at 05:14:25PM -0700, Ian Clarke wrote:
> On Thu, Sep 05, 2002 at 05:06:48PM -0700, Reuben Balik wrote:
> > I was looking at the Freenet project page and I
> > noticed that there hasn't been a release until
> > December. Since the latest build seems pretty stable
> > (I
L:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/de09fe03/attachment.pgp>
>From Ian Clarke
>I just experienced an out-of-memory error - could people check their
>recently upgraded nodes and make sure that memory usage is sane?
>
>Perhaps time for another round of heap-dumps.
My new build 498 node runs out of a default-sized heap within hours, and doing
-Xmx256m fills
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/3295a528/attachment.pgp>
bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/8a0fcb96/attachment.pgp>
I was looking at the Freenet project page and I
noticed that there hasn't been a release until
December. Since the latest build seems pretty stable
(I haven't had a problem) and has a nice interface
(although I dislike the Internet Explorer Sucks
message; couldn't you just put a small warning on
Sorry about the previous message -somebodies SW being too clever and
guessing what I wanted...
I've been getting out of memory crashes too: I've also noticed that the
ticker contents display is filling up with megabytes of:
freenet.node.states.announcing.Announcing$ScheduleAnnouncing at
On Thu, Sep 05, 2002 at 12:52:46PM +0100, Matthew Toseland wrote:
> On Wed, Sep 04, 2002 at 11:19:39PM -0700, Ian Clarke wrote:
> > For those too lazy to type "ant javadoc", here are the current Freenet
> > javadocs:
> >
> > http://hawk.freenetproject.org/~ian/freenet_javadocs/api/
> >
> >
An HTML attachment was scrubbed...
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/0656dae4/attachment.html>
: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/cb09fc58/attachment.pgp>
Either we're reporting it now, or something is using a cubic fuckton
more RAM. I've got the VM limit set at 96 meg, and I'm sorry, but that's
"More Then Enough" to do the job. I've only got a half gig on this
box, so running to 128meg is going to start hurting (especially if there's
a
My bad, that's what I get for testing on small files ^_^. (when your
files are only 1meg, you don't hit hex, and hence, you know, it's all good
then :-p.). sorry!
Anyhow, looking at GJ's patch, I understand what I did that was
incompatable now, so should I fix it to be compatable with the old
zer Inc. http://www.uprizer.com/
Personal Homepage http://locut.us/
-- next part --
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/4caff113/attachment.pgp>
<>
> The problem with make is javac has a stupid startup time, so is best called
> with 30+ files at
> once. Make likes to call things sequentially. ant is designed to call javac
> once with
> all the filenames.
That's why I use Jikes :)
___
devl
urceforge.net
Freenet/Coldstore open source hacker.
Looking for $coding (I'm cheap)
-- next part --
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/e9c79cf1/attachment.pgp>
er.
Looking for $coding (I'm cheap)
-- next part --
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/3d69a77f/attachment.pgp>
nt was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/1ccd1a24/attachment.pgp>
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/c023f705/attachment.pgp>
next part --
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 232 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/49e5ec17/attachment.pgp>
2 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/be5c8dab/attachment.pgp>
fish writes:
>
> My bad, that's what I get for testing on small files ^_^. (when your
> files are only 1meg, you don't hit hex, and hence, you know, it's all good
> then :-p.). sorry!
>
> Anyhow, looking at GJ's patch, I understand what I did that was
> incompatable now, so should I fix it to
Ian Clarke writes:
>
> --S1BNGpv0yoYahz37
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
>
> Well, in theory - yes, in practice, it ain't so easy.
>
> The NodeInfoServlet resides on port 8890, that is the thing that
Ian Clarke writes:
>
> --iFRdW5/EC4oqxDHL
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
>
> Does this mean that the "view page source" link that comes up when the=20
> anonymity doesn't work in IE?
Yes. And the safest
for $coding (I'm cheap)
-- next part --
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/79629e3e/attachment.pgp>
achment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL:
<https://emu.freenetproject.org/pipermail/devl/attachments/20020905/ea4e6de5/attachment.pgp>
Matthew Toseland writes:
>
> --6TrnltStXW4iwmi0
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
>
> On Thu, Aug 29, 2002 at 11:59:50AM -0700, Ian Clarke wrote:
> > I would like to encourage those still using the Makefile
Just a rule of thumb - I suggest that we avoid upping the build number
more than once a week.
Ian.
--
Ian Clarkeian at freenetproject.org
Founder & Coordinator, The Freenet Projecthttp://freenetproject.org/
Chief Technology Officer, Uprizer Inc.
Ian Clarke writes:
>
> --gKMricLos+KVdGMg
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: inline
> Content-Transfer-Encoding: quoted-printable
>
> Christ, no matter what horrible mutilations of the code I wilfully admit=20
> on this mailing list, I still fail to provoke a
On Wed, Sep 04, 2002 at 11:12:03PM -0700, Tracy R Reed wrote:
> Next time my datastore tosses its cookies what should I do? What
> information do you need? I'd like to help however possible.
I don't think there is much that can be done by the general public, I
suspect that diagnosing the problem
For those too lazy to type "ant javadoc", here are the current Freenet
javadocs:
http://hawk.freenetproject.org/~ian/freenet_javadocs/api/
These are particularly useful for figuring out how stuff like the
filesystem implementation (hint hint) work.
Ian.
--
Ian Clarke
On Wed, Sep 04, 2002 at 06:52:40PM -0700, Ian Clarke spake thusly:
> It his high-time we took concerted action to kill the datastore bug.
Glory glory hallelujah!
Next time my datastore tosses its cookies what should I do? What
information do you need? I'd like to help however possible.
> -
Just a rule of thumb - I suggest that we avoid upping the build number
more than once a week.
Ian.
--
Ian Clarke[EMAIL PROTECTED]
Founder Coordinator, The Freenet Projecthttp://freenetproject.org/
Chief Technology Officer, Uprizer Inc.
Ian Clarke writes:
--S1BNGpv0yoYahz37
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Well, in theory - yes, in practice, it ain't so easy.
The NodeInfoServlet resides on port 8890, that is the thing that draws the =
On Wed, Sep 04, 2002 at 11:44:52PM -0700, Ian Clarke wrote:
Just a rule of thumb - I suggest that we avoid upping the build number
more than once a week.
I suggest we only up the version number when there is a change to the
externally (to the network) visible behaviour of the node, and we do
On Thu, Sep 05, 2002 at 03:02:06AM -0400, Dan Merillat wrote:
Ian Clarke writes:
--iFRdW5/EC4oqxDHL
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
Does this mean that the view page source link that comes up
The problem with make is javac has a stupid startup time, so is best called with 30+
files at
once. Make likes to call things sequentially. ant is designed to call javac once
with
all the filenames.
That's why I use Jikes :)
___
devl mailing
On Thu, Sep 05, 2002 at 12:52:46PM +0100, Matthew Toseland wrote:
On Wed, Sep 04, 2002 at 11:19:39PM -0700, Ian Clarke wrote:
For those too lazy to type ant javadoc, here are the current Freenet
javadocs:
http://hawk.freenetproject.org/~ian/freenet_javadocs/api/
These are
On Thu, Sep 05, 2002 at 12:51:47PM +0100, Matthew Toseland wrote:
On Wed, Sep 04, 2002 at 11:44:52PM -0700, Ian Clarke wrote:
Just a rule of thumb - I suggest that we avoid upping the build number
more than once a week.
I suggest we only up the version number when there is a change to the
Some component is leaking memory. I have been suspecting that for some
time. What we need is for somebody to run some heap profiles on long
running nodes.
See: http://www.javaworld.com/javaworld/jw-12-2001/jw-1207-hprof.html
On Thu, Sep 05, 2002 at 04:05:42PM -0400, Dan Merillat wrote:
On Thu, Sep 05, 2002 at 01:54:07PM -0700, Ian Clarke wrote:
So - question - do OutputStreams/Writers to Servlets (as obtained from
HttpServletResponse objects) need to be closed before the doGet() method
of the servlet terminates?
resp.flushBuffer() should be called.
--
Oskar Sandberg
From Ian Clarke [EMAIL PROTECTED]
I just experienced an out-of-memory error - could people check their
recently upgraded nodes and make sure that memory usage is sane?
Perhaps time for another round of heap-dumps.
My new build 498 node runs out of a default-sized heap within hours, and doing
On Thu, Sep 05, 2002 at 06:49:59PM -0400, Benjamin Coates wrote:
My new build 498 node runs out of a default-sized heap within hours, and doing
-Xmx256m fills up and starts thrashing on my 128MB system.
how do you get a heap dump?
Something like this:
java
I've been getting out of memory crashes too: I've also noticed that the ticker contents displayis filling up with megabytes of:
freenet.node.states.announcing.Announcing$ScheduleAnnouncing@e90e23
could this have something to do with it ?
The log looks like this:
Snip
Sep 5, 2002 2:26:47 AM
Sorry about the previous message -somebodies SW being too clever and
guessing what I wanted...
I've been getting out of memory crashes too: I've also noticed that the
ticker contents display is filling up with megabytes of:
freenet.node.states.announcing.Announcing$ScheduleAnnouncing@959352
I was looking at the Freenet project page and I
noticed that there hasn't been a release until
December. Since the latest build seems pretty stable
(I haven't had a problem) and has a nice interface
(although I dislike the Internet Explorer Sucks
message; couldn't you just put a small warning on
On Thu, Sep 05, 2002 at 05:06:48PM -0700, Reuben Balik wrote:
I was looking at the Freenet project page and I
noticed that there hasn't been a release until
December. Since the latest build seems pretty stable
(I haven't had a problem) and has a nice interface
(although I dislike the
On Thu, Sep 05, 2002 at 04:53:19PM -0700, Simon G wrote:
I've been getting out of memory crashes too: I've also noticed that the
ticker contents display is filling up with megabytes of:
freenet.node.states.announcing.Announcing$ScheduleAnnouncing@959352
This could easily be the problem as I
05-Sep-02 6:37:37 PM (freenet.node.Node, main): Starting interfaces..
05-Sep-02 7:59:05 PM (freenet.client.http.FproxyServlet, QThread-321:
freenet.interfaces.LocalInterface$ConnectionShell@1531d48 (time: 1031270280)): Error
sending data to browser: java.io.IOException: Broken pipe
05-Sep-02
On Thu, 5 Sep 2002 20:38:56 -0400
Greg Wooledge [EMAIL PROTECTED] wrote:
[DSB occuring right after Upstream node sent bad data!]
I've seen this too, but I've also seen the bad data message without
any obvious DSB consequences.
I've also gotten messages around that time about unable to send
On Thu, Sep 05, 2002 at 09:23:58PM -0500, Travis Bemann wrote:
But still, we should be extremely firm in warning users about the use
of M$IE and the insecurity of it. I personally think that even
ALLOWING users to view fproxy beyond a rejection page with links to
other browsers one can get
On Thu, Sep 05, 2002 at 08:38:56PM -0400, Greg Wooledge wrote:
(The rumors of memory leaks in the
newest builds scare me, especially since I came back this afternoon
to find my node using 253 out of 256 MB allocated memory, so I reverted.)
Matthew checked in a potential fix for these memory
On Thu, Sep 05, 2002 at 10:21:48PM -0700, Ian Clarke wrote:
On Thu, Sep 05, 2002 at 09:23:58PM -0500, Travis Bemann wrote:
But still, we should be extremely firm in warning users about the use
of M$IE and the insecurity of it. I personally think that even
ALLOWING users to view fproxy
54 matches
Mail list logo
