You are viewing a plain text version of this content. The canonical link for it is here.
Posted to kato-spec@incubator.apache.org by Steve Poole <sp...@googlemail.com> on 2009/11/17 10:46:37 UTC
[TOPIC] Snapshot data collection definition - dump triggers
Andreas posted some interesting information here
http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D5024AE296@DYNASERV.dynatrace.local%3E
This thread is to discuss the topic of how and when snapshot dumps are
triggered.
Andreas has started with asking for
1) scheduled (timed) dumps
2) event driven dumps including fatal JVM crashed type events
Question: How adaptive would this timed dump have to be? I can imagine
having a regular heart beat dump but would we want to be able to modify the
timeing on the fly?
Question: Would we want to have multiple timers running with different
dump output?
Question: What are the top events that we need to consider should be
covered - there is the usual OOM but what else?
--
Steve
Re: [TOPIC] Snapshot data collection definition - dump triggers
Posted by Sonal Goyal <so...@gmail.com>.
Hi Steve,
I would suspect that the VM would have a handle to the native stack and thus
be able to signal if something goes wrong. Is there something I am missing
here?
Thanks and Regards,
Sonal
On Fri, Nov 20, 2009 at 4:20 PM, Steve Poole <sp...@googlemail.com>wrote:
> On Thu, Nov 19, 2009 at 11:37 AM, Sonal Goyal <so...@gmail.com>
> wrote:
>
> > Hi Steve,
> >
> > Do we see a scenario where different timers will be capturing different
> > data
> > ? For example, simple dump of class instances every nth interval, and
> > detailed dump every mth interval?
> >
> > Hmm - yes I suspect you're right - that does sound likely.
>
>
> > For event driven dumps, we may like to look at JNI related crashes.
> >
> > Agree - we do we want to do that. I don't know if there is any way of
> specifically noticing a JNI crash though.
> It would be great to do something like Dump.setOnJNICrash(true) but I
> don't know if we can actually seperate out
> the event from just a straightforward dump request.
>
> Thanks and Regards,
> > Sonal
> >
> >
> > On Tue, Nov 17, 2009 at 4:16 PM, Steve Poole <spoole167@googlemail.com
> > >wrote:
> >
> > > Andreas posted some interesting information here
> > >
> >
> http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D50data
> > > in the24AE296@DYNASERV.dynatrace.local%3E<
> >
> http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D5024AE296@DYNASERV.dynatrace.local%3E
> > >
> > >
> > > This thread is to discuss the topic of how and when snapshot dumps
> are
> > > triggered.
> > >
> > > Andreas has started with asking for
> > >
> > > 1) scheduled (timed) dumps
> > > 2) event driven dumps including fatal JVM crashed type events
> > >
> > > Question: How adaptive would this timed dump have to be? I can
> > imagine
> > > having a regular heart beat dump but would we want to be able to modify
> > the
> > > timeing on the fly?
> > >
> > > Question: Would we want to have multiple timers running with
> different
> > > dump output?
> > >
> > > Question: What are the top events that we need to consider should be
> > > covered - there is the usual OOM but what else?
> > >
> > > --
> > > Steve
> > >
> >
>
>
>
> --
> Steve
>
Re: [TOPIC] Snapshot data collection definition - dump triggers
Posted by Steve Poole <sp...@googlemail.com>.
On Thu, Nov 19, 2009 at 11:37 AM, Sonal Goyal <so...@gmail.com> wrote:
> Hi Steve,
>
> Do we see a scenario where different timers will be capturing different
> data
> ? For example, simple dump of class instances every nth interval, and
> detailed dump every mth interval?
>
> Hmm - yes I suspect you're right - that does sound likely.
> For event driven dumps, we may like to look at JNI related crashes.
>
> Agree - we do we want to do that. I don't know if there is any way of
specifically noticing a JNI crash though.
It would be great to do something like Dump.setOnJNICrash(true) but I
don't know if we can actually seperate out
the event from just a straightforward dump request.
Thanks and Regards,
> Sonal
>
>
> On Tue, Nov 17, 2009 at 4:16 PM, Steve Poole <spoole167@googlemail.com
> >wrote:
>
> > Andreas posted some interesting information here
> >
> http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D50data
> > in the24AE296@DYNASERV.dynatrace.local%3E<
> http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D5024AE296@DYNASERV.dynatrace.local%3E
> >
> >
> > This thread is to discuss the topic of how and when snapshot dumps are
> > triggered.
> >
> > Andreas has started with asking for
> >
> > 1) scheduled (timed) dumps
> > 2) event driven dumps including fatal JVM crashed type events
> >
> > Question: How adaptive would this timed dump have to be? I can
> imagine
> > having a regular heart beat dump but would we want to be able to modify
> the
> > timeing on the fly?
> >
> > Question: Would we want to have multiple timers running with different
> > dump output?
> >
> > Question: What are the top events that we need to consider should be
> > covered - there is the usual OOM but what else?
> >
> > --
> > Steve
> >
>
--
Steve
Re: [TOPIC] Snapshot data collection definition - dump triggers
Posted by Sonal Goyal <so...@gmail.com>.
Hi Steve,
Do we see a scenario where different timers will be capturing different data
? For example, simple dump of class instances every nth interval, and
detailed dump every mth interval?
For event driven dumps, we may like to look at JNI related crashes.
Thanks and Regards,
Sonal
On Tue, Nov 17, 2009 at 4:16 PM, Steve Poole <sp...@googlemail.com>wrote:
> Andreas posted some interesting information here
> http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D50data
> in the24AE296@DYNASERV.dynatrace.local%3E<http://mail-archives.apache.org/mod_mbox/incubator-kato-spec/200911.mbox/%3C6EB74B946FF4CF4BAA39332269A659D5024AE296@DYNASERV.dynatrace.local%3E>
>
> This thread is to discuss the topic of how and when snapshot dumps are
> triggered.
>
> Andreas has started with asking for
>
> 1) scheduled (timed) dumps
> 2) event driven dumps including fatal JVM crashed type events
>
> Question: How adaptive would this timed dump have to be? I can imagine
> having a regular heart beat dump but would we want to be able to modify the
> timeing on the fly?
>
> Question: Would we want to have multiple timers running with different
> dump output?
>
> Question: What are the top events that we need to consider should be
> covered - there is the usual OOM but what else?
>
> --
> Steve
>