RFR : 8225345: Provide Cloud IAAS related info on Linux in the hs_err file
Kodewerk
kirk at kodewerk.com
Fri Jun 7 16:18:26 UTC 2019
Hi Thomas,
I have to agree with you. As it stands, the hs_err does not contain any information that cannot be passed to me with the need of an NDA. Once you start digging deeper into the end users environment you start raising serious security concerns. I would suggest that this information should be collectable in anger and not by default.
Kind regards,
Kirk
> On Jun 7, 2019, at 7:08 AM, Thomas Stüfe <thomas.stuefe at gmail.com> wrote:
>
> On Fri, Jun 7, 2019 at 3:27 PM Baesken, Matthias <matthias.baesken at sap.com>
> wrote:
>
>>> fact the machine has an AMI BIOS is not.
>>
>> Printing the BIOS was not part of my original patch .
>>
>> Saying - "Printing the BIOS is not relevant " ... we had a few very
>> complicated and time consuming JVM support cases in the past where
>> BIOS updates where involved .
>> ( but that’s another discussion )
>>
>
> So where do you draw the line? Since you obviously do not draw it here. At
> clear user names? Content of user home directory? Topology of the internal
> network? All of these may be useful to some support case. Which, to me,
> does not mean they have a place in a crash report.
>
> Convenience for developer support is only one side. User trust is another.
> And there, it does not matter if you think this is harmless. It matters
> what users think. The moment the first blogger writes about "Look what I
> found in a JDK crash report" it can be damaging to the platform as a whole.
>
> --Thomas
>
>
>>
>> Best regards , Matthias
>>
>>
>>
>>> -----Original Message-----
>>> From: David Holmes <david.holmes at oracle.com>
>>> Sent: Freitag, 7. Juni 2019 15:10
>>> To: Baesken, Matthias <matthias.baesken at sap.com>; Thomas Stüfe
>>> <thomas.stuefe at gmail.com>
>>> Cc: hotspot-dev at openjdk.java.net
>>> Subject: Re: RFR : 8225345: Provide Cloud IAAS related info on Linux in
>> the
>>> hs_err file
>>>
>>> On 7/06/2019 10:26 pm, Baesken, Matthias wrote:
>>>> The last iteration of the discussion was just to write out the dmi
>>>> information , without interpretation.
>>>>
>>>> These are technical values .
>>>
>>> No they are not. They still convey the same information that we're
>>> concerned about.
>>>
>>> Look at it this way, we could have been putting SMBIOS/DMI information
>>> in the hs_err report from day one. We don't. It's never been relevant.
>>> The fact we're executing on an x86 family 5 cpu may be relevant. The
>>> fact the machine has an AMI BIOS is not.
>>>
>>> David
>>> -----
>>>
>>>> Regards, Matthias
>>>>
>>>> *From:*Thomas Stüfe <thomas.stuefe at gmail.com>
>>>> *Sent:* Freitag, 7. Juni 2019 13:58
>>>> *To:* David Holmes <david.holmes at oracle.com>; Baesken, Matthias
>>>> <matthias.baesken at sap.com>
>>>> *Cc:* hotspot-dev at openjdk.java.net
>>>> *Subject:* Re: RFR : 8225345: Provide Cloud IAAS related info on Linux
>>>> in the hs_err file
>>>>
>>>> On Thu, Jun 6, 2019 at 2:51 PM David Holmes <david.holmes at oracle.com
>>>> <mailto:david.holmes at oracle.com>> wrote:
>>>>
>>>> On 6/06/2019 10:17 pm, Baesken, Matthias wrote:
>>>>> Hi David , sure I can guard this by
>> "ExtensiveErrorReports" .
>>>>>
>>>>>>
>>>>>> Sorry but how is it at all relevant to crash diagnosis?
>>>>>>
>>>>>
>>>>> It gives a more complete picture where you run , helps to
>>>> improve the error analysis .
>>>>> ( like - involve the Cloud escalation team because ... )
>>>>
>>>> Sorry not seeing it.
>>>>
>>>>>>
>>>>>> And maybe a customer doesn't want you to know whose Cloud
>>>> services they use!
>>>>>>
>>>>>
>>>>> Some people might not want the number of cores in hs_err, some
>>>> might not want to OS version ,
>>>>> some people might not want something else in hs_err .
>>>>> That's always hard to say , but I think the Cloud info is on
>> the
>>>> same technical level than what we currently already have in the
>>>> hs_err so
>>>>> Nothing special .
>>>>
>>>> Vastly different IMO. Whose Cloud services company X uses could be
>>>> considered commercial-in-confidence. I don't see anything
>>>> "technical" in
>>>> knowing whose cloud was used.
>>>>
>>>> Hate to say this, but I agree with David. This is even making me
>>>> uncomfortable.
>>>>
>>>> To say this clearly, I am not denying this information is not useful.
>> Of
>>>> course it is. But I think putting this into crash reports is a) not the
>>>> right place since it has not much to do with crashes and b) has the
>>>> potential of damaging the trust of users.
>>>>
>>>> More generally, I would like a clear conversation and negotiation -
>> with
>>>> written down results - about what should and should not go into the
>>>> crash report. Otherwise this is will be a continuous source of
>> friction.
>>>> I value an easy and trustful cooperation and trustful users more highly
>>>> that detailed crash reports (which I do value very highly! :-).
>>>>
>>>> As an added benefit of such a discussion, having a Wiki page defining
>>>> which information a crash report may contain (and which not) can build
>>>> trust with our users.
>>>>
>>>> Just my five cent,
>>>>
>>>> Thomas
>>>>
>>>> David
>>>> -----
>>>>
>>>>>
>>>>> Best regards, Matthias
>>>>>
>>>>>
>>>>>> From: David Holmes <david.holmes at oracle.com
>>>> <mailto:david.holmes at oracle.com>>
>>>>>> Sent: Donnerstag, 6. Juni 2019 13:51
>>>>>> To: Baesken, Matthias <matthias.baesken at sap.com
>>>> <mailto:matthias.baesken at sap.com>>; 'hotspot-
>>>>>> dev at openjdk.java.net <mailto:dev at openjdk.java.net>'
>>>> <hotspot-dev at openjdk.java.net <mailto:hotspot-
>>> dev at openjdk.java.net>>
>>>>>> Subject: Re: RFR : 8225345: Provide Cloud IAAS related info on
>>>> Linux in the
>>>>>> hs_err file
>>>>>>
>>>>>> Hi Matthias,
>>>>>>
>>>>>> On 6/06/2019 9:29 pm, Baesken, Matthias wrote:
>>>>>>> Hello, please review the following change .
>>>>>>> It enhances the hs_error file output on Linux by Cloud IAAS
>>>> detection .
>>>>>>>
>>>>>>> The DMI information exposed by the /sys virtual filesystem
>>>> can be used
>>>>>> to detect various Cloud IAAS on Linux.
>>>>>>> It is helpful, for instance in support cases, to get some
>>>> information about
>>>>>> the Cloud IAAS in the hs_error file .
>>>>>>
>>>>>> Sorry but how is it at all relevant to crash diagnosis?
>>>>>>
>>>>>> And maybe a customer doesn't want you to know whose Cloud
>>>> services they
>>>>>> use!
>>>>>>
>>>>>> This seems like something that should be guarded by
>>>>>> ExtensiveErrorReports - at best.
>>>>>>
>>>>>> David
>>>>>> -----
>>>>>>
>>>>>>>
>>>>>>> Bug/webrev :
>>>>>>>
>>>>>>> https://bugs.openjdk.java.net/browse/JDK-8225345
>>>>>>>
>>>>>>> http://cr.openjdk.java.net/~mbaesken/webrevs/8225345.0/
>>>>>>>
>>>>>>>
>>>>>>> Thanks, Matthias
>>>>>>>
>>>>
>>
More information about the hotspot-dev
mailing list