The config DB seems to have the answers and this could/should obviously just
be reused. The SystemID is generated using Data::UUID which promises it to
be unique. As I see it we have our data in the database.
# config show sysconfig
sysconfig=configuration
InstallEpoch=1372281536
KeyboardType=pc
Keytable=dk
Language=en_US
PreviousSystemMode=serveronly
Registration=none
ReleaseVersion=8.1
SystemID=d0a6d2e7-f4be-4ebf-ae5a-7507775690ab
From: [email protected]
[mailto:[email protected]] On Behalf Of Hsing-Foo Wang
Sent: 10. marts 2014 15:40
To: [email protected]
Subject: Re: [discussion] Install statistics for SME Server [SMOLT]
I hope Shad would be willing to speak up, for he is fully aware and
maintained all 'phone home' and install UUID, EPOCH stuff. Hopefully it will
prevent us from re-inventing the wheel.
On 10-3-2014 15:22, Jesper Knudsen wrote:
I agree that a unique ID would be beneficial in the long run and would also
eliminate discussions on polluted data - I was only trying to make a point
that we do not need an opt-in for the suggested data.
I have used the MAC address before for that purpose - before VMs it was
unique but now we need to combined with something more. Maybe the UUID for
the root device (found via blkid).
f.ex:
/dev/mapper/main-root: UUID="a41a9845-53be-489b-969c-bee4d387af3d"
TYPE="ext3"
From: [email protected]
[mailto:[email protected]] On Behalf Of Jonathan Martens
Sent: 10. marts 2014 09:58
To: [email protected]
Subject: Re: [discussion] Install statistics for SME Server [SMOLT]
On 9-3-2014 23:33, Jesper Knudsen wrote:
The only think that this would "reveal" beyond what a YUM update does is a
unique ID and install date (why do we need that?). I do not think we need an
opt-in for that. Basically, as long as the data gathering is so simple I do
not even think a unique ID is needed - we are not looking for exact science
here but marketing number and if a few servers are behind the same router/IP
that will, or should not, disturb the picture much.
We could and should properly make an opt-in solution if we decided to gather
more detailed (and usefull??) data as I suggested.
Greetings,
Jesper
Although a unique key does not sound as a requirement for you, I know it
makes extension to the data model much easier, I think it is not such a big
hassle to create one for every server. I could imagine that in the future we
would like to now the install base on a per version base and would like to
have an update when a machine changes version number for instance, this
would be hard to do without a unique id. In other words without the unique
ID you do not know if you get duplicates and statistics data is polluted
easily, which IMHO defeats the purpose of this exercise.
Kind regards,
Jonathan
_______________________________________________
Discussion about project organisation and overall direction
To unsubscribe, e-mail [email protected]
Searchable archive at http://lists.contribs.org/mailman/public/discussion/
_______________________________________________
Discussion about project organisation and overall direction
To unsubscribe, e-mail [email protected]
Searchable archive at http://lists.contribs.org/mailman/public/discussion/
