dhis to submit anonymous data

Hi there,

during the Uganda workshop last month we had an interesting suggestion from one of the participants.

The idea is to let DHIS submit anonymous, non-sensitive data to a central system in order to gather basic statistics related to usage. The data sent would include:

  • A random, anonymous system identifier

  • DHIS 2 version

  • Java server version

  • Operating system and version

Then we could make an optional feature for “activating” or “registering” your DHIS instance, where you could optionally inform about:

  • System contact person

  • System description

It would be quite useful and interesting to know things like:

  • How many DHIS instances there are out there

  • How fast are new DHIS versions adopted

  • What is the dominant server operating system/environment

This info would of course be disclosed only in aggregate form.

What do people think about this? Any strong objections?

best regards,

Lars

Hi Lars,

Sounds interesting. Wish I would have been there for the discussion. Sounds like it might be possible, assuming the process and data is transparent, and allow for an option to “opt-out” of collection of such data.

Having said that, I think we should be exceedingly careful. I assume that the data would be beamed back to a central server someplace. Who would control that server? Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

Regards,

Jason

···

On Thu, Jun 27, 2013 at 10:07 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi there,

during the Uganda workshop last month we had an interesting suggestion from one of the participants.

The idea is to let DHIS submit anonymous, non-sensitive data to a central system in order to gather basic statistics related to usage. The data sent would include:

  • A random, anonymous system identifier
  • DHIS 2 version
  • Java server version
  • Operating system and version

Then we could make an optional feature for “activating” or “registering” your DHIS instance, where you could optionally inform about:

  • System contact person
  • System description

It would be quite useful and interesting to know things like:

  • How many DHIS instances there are out there
  • How fast are new DHIS versions adopted
  • What is the dominant server operating system/environment

This info would of course be disclosed only in aggregate form.

What do people think about this? Any strong objections?

best regards,

Lars


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Hi Lars and Jason,

Good idea, and good questions about security.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Cheers,
Jim

···

On Thu, Jun 27, 2013 at 10:07 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi there,

during the Uganda workshop last month we had an interesting suggestion from one of the participants.

The idea is to let DHIS submit anonymous, non-sensitive data to a central system in order to gather basic statistics related to usage. The data sent would include:

  • A random, anonymous system identifier
  • DHIS 2 version
  • Java server version
  • Operating system and version

Then we could make an optional feature for “activating” or “registering” your DHIS instance, where you could optionally inform about:

  • System contact person
  • System description

It would be quite useful and interesting to know things like:

  • How many DHIS instances there are out there
  • How fast are new DHIS versions adopted
  • What is the dominant server operating system/environment

This info would of course be disclosed only in aggregate form.

What do people think about this? Any strong objections?

best regards,

Lars


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

Hi,

thanks for the good questions and comments.

···

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason

···

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

I agree with Jason on all counts so won’t repeat them.

A mandatory “ET call home” feature would not and should not be generally acceptable. Besides which its not too clear how, or more pertinently, when, this exchange will happen. On first boot? Every boot. Periodically?

Anyway I don’t like it and also worry that these things have a nasty habit of leading to scope creep.

I think the idea of periodically scanning for releases and updates is good. Also planning to integrate such behaviour in dhis2-tools.

Bob

···

On 28 June 2013 15:01, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

Okay thanks for feedback.

To clarify, for the “calling home” part we will only collect the DHIS 2 version (no IP, no Java version etc). In other words we will only register that there exists a DHIS version out there with a given version. It will not be possible to track it back to the IP.

Then, what confuses me is; how could the fact that a government uses DHIS 2 as their health system possibly be sensitive information or violate country laws? It will already be a public Internet facing system, which could be revealed through a google search; and with likely thousands of users, knowing of its existence.

···

On Fri, Jun 28, 2013 at 5:24 PM, Bob Jolliffe bobjolliffe@gmail.com wrote:

I agree with Jason on all counts so won’t repeat them.

A mandatory “ET call home” feature would not and should not be generally acceptable. Besides which its not too clear how, or more pertinently, when, this exchange will happen. On first boot? Every boot. Periodically?

Anyway I don’t like it and also worry that these things have a nasty habit of leading to scope creep.

I think the idea of periodically scanning for releases and updates is good. Also planning to integrate such behaviour in dhis2-tools.

Bob

On 28 June 2013 15:01, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

Perhaps it could be modeled after the Drupal Update Stats module. This module generates simple stats that can be seen here. https://drupal.org/project/usage/drupal you can read more about it here. https://drupal.org/node/329620

I’d also support a slightly more detailed optional submission as well that does send OS, Java version, Web Server and database server, though rather than it being a service that runs in the background there would be a submit button on the About DHIS2 page. So it would be very clear what information was being submitted and when that information was being submitted. This information would be very help for the development team to have a clear idea of the target DHIS2 environments and make sure they are properly supported.

Dan

···

On Fri, Jun 28, 2013 at 5:24 PM, Bob Jolliffe bobjolliffe@gmail.com wrote:

I agree with Jason on all counts so won’t repeat them.

A mandatory “ET call home” feature would not and should not be generally acceptable. Besides which its not too clear how, or more pertinently, when, this exchange will happen. On first boot? Every boot. Periodically?

Anyway I don’t like it and also worry that these things have a nasty habit of leading to scope creep.

I think the idea of periodically scanning for releases and updates is good. Also planning to integrate such behaviour in dhis2-tools.

Bob

On 28 June 2013 15:01, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

Agree with Dan that a submit button on the about page (which previewed the information that would be sent) is better than ET calling home.

Though we need to remain clear about the usage distinction of sharing with developers and feeding a more public repo. Need to think a bit more.

···

On 28 June 2013 17:10, Dan Cocos dan@dancocos.com wrote:

Perhaps it could be modeled after the Drupal Update Stats module. This module generates simple stats that can be seen here. https://drupal.org/project/usage/drupal you can read more about it here. https://drupal.org/node/329620

I’d also support a slightly more detailed optional submission as well that does send OS, Java version, Web Server and database server, though rather than it being a service that runs in the background there would be a submit button on the About DHIS2 page. So it would be very clear what information was being submitted and when that information was being submitted. This information would be very help for the development team to have a clear idea of the target DHIS2 environments and make sure they are properly supported.

Dan

On Jun 28, 2013, at 11:54 AM, Lars Helge Øverland larshelge@gmail.com wrote:

Okay thanks for feedback.

To clarify, for the “calling home” part we will only collect the DHIS 2 version (no IP, no Java version etc). In other words we will only register that there exists a DHIS version out there with a given version. It will not be possible to track it back to the IP.

Then, what confuses me is; how could the fact that a government uses DHIS 2 as their health system possibly be sensitive information or violate country laws? It will already be a public Internet facing system, which could be revealed through a google search; and with likely thousands of users, knowing of its existence.


Mailing list: https://launchpad.net/~dhis2-devs
Post to : dhis2-devs@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-devs
More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 5:24 PM, Bob Jolliffe bobjolliffe@gmail.com wrote:

I agree with Jason on all counts so won’t repeat them.

A mandatory “ET call home” feature would not and should not be generally acceptable. Besides which its not too clear how, or more pertinently, when, this exchange will happen. On first boot? Every boot. Periodically?

Anyway I don’t like it and also worry that these things have a nasty habit of leading to scope creep.

I think the idea of periodically scanning for releases and updates is good. Also planning to integrate such behaviour in dhis2-tools.

Bob

On 28 June 2013 15:01, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars

Hi Lars,
What I mean by the IP being collected, is it “might” be collected (unless you are careful) by your reverse proxy or upstream providers. Even though whatever protocol you might come up with might not collect it, it would require you to purposefully NOT collect it though unintentional means, such as through the logs of your upstream proxies or by your ISP. Additionally, I think it needs to be crystal clear who the “DHIS core team” is, under what laws they operate, and how it can be certified that such information is not being collected (other than just saying that it is not being done).

I think the fact that many ministries and governments use DHIS2 is particularly relevant in this case, because I have run up against a lot of distrust of the whole cloud computing model which many implementers (including myself) are using. Adding the possible collection of any data by the “DHIS core team” to this, could further complicate already long negotiations on the issue of hosting. Although the data which DHIS2 collects is generally not regarded to be secret, there have been instances of which I personally know of, in which this is the case. Since you have stated the information would be open-source, it would mean that potentially it could be used for research purposes, or other unknown (and possibly nefarious) purposes, which again, could complicate already sensitive discussions with some of the DHIS2 users.

So, I think until you make clear exactly what is going to done with the data, who will manage and protect it, I think the “opt in” model would be the only really feasible option, and even then, make it exactly clear what is going to be submitted and to whom and for what purpose.

Regards,

Jason

···

On Fri, Jun 28, 2013 at 6:21 PM, Bob Jolliffe bobjolliffe@gmail.com wrote:

Agree with Dan that a submit button on the about page (which previewed the information that would be sent) is better than ET calling home.

Though we need to remain clear about the usage distinction of sharing with developers and feeding a more public repo. Need to think a bit more.


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On 28 June 2013 17:10, Dan Cocos dan@dancocos.com wrote:

Perhaps it could be modeled after the Drupal Update Stats module. This module generates simple stats that can be seen here. https://drupal.org/project/usage/drupal you can read more about it here. https://drupal.org/node/329620

I’d also support a slightly more detailed optional submission as well that does send OS, Java version, Web Server and database server, though rather than it being a service that runs in the background there would be a submit button on the About DHIS2 page. So it would be very clear what information was being submitted and when that information was being submitted. This information would be very help for the development team to have a clear idea of the target DHIS2 environments and make sure they are properly supported.

Dan

On Jun 28, 2013, at 11:54 AM, Lars Helge Øverland larshelge@gmail.com wrote:

Okay thanks for feedback.

To clarify, for the “calling home” part we will only collect the DHIS 2 version (no IP, no Java version etc). In other words we will only register that there exists a DHIS version out there with a given version. It will not be possible to track it back to the IP.

Then, what confuses me is; how could the fact that a government uses DHIS 2 as their health system possibly be sensitive information or violate country laws? It will already be a public Internet facing system, which could be revealed through a google search; and with likely thousands of users, knowing of its existence.


Mailing list: https://launchpad.net/~dhis2-devs
Post to : dhis2-devs@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-devs
More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 5:24 PM, Bob Jolliffe bobjolliffe@gmail.com wrote:

I agree with Jason on all counts so won’t repeat them.

A mandatory “ET call home” feature would not and should not be generally acceptable. Besides which its not too clear how, or more pertinently, when, this exchange will happen. On first boot? Every boot. Periodically?

Anyway I don’t like it and also worry that these things have a nasty habit of leading to scope creep.

I think the idea of periodically scanning for releases and updates is good. Also planning to integrate such behaviour in dhis2-tools.

Bob

On 28 June 2013 15:01, Jason Pickering jason.p.pickering@gmail.com wrote:

Hi Lars,

I think it is important to get it right from the beginning. following the lead of other big open source projects. It is critical to remember that many of the organisations using DHIS2 are governments, and there could be some possible sensitivies about the DHIS core team collecting any sort of information.

Before you go too far with this, I would think it would be good to identify what the purpose (which you mention) of collecting any data would be, and spell it out very clearly, and why opting out would not be an option. I would also be clear about all of the legal technicalities, including what county’s law would regulate the collection of such information.

Even if you were not to collect the IP’s, there is of course always the possibility that they could be collected upstream. This might not be directly the core team’s worry, but it might be.

Regards,

Jason


Mailing list: https://launchpad.net/~dhis2-users

Post to : dhis2-users@lists.launchpad.net

Unsubscribe : https://launchpad.net/~dhis2-users

More help : https://help.launchpad.net/ListHelp

On Fri, Jun 28, 2013 at 2:35 PM, Lars Helge Øverland larshelge@gmail.com wrote:

Hi,

thanks for the good questions and comments.

This information will be stored in some central system. The system will be managed by the DHIS core team. That system will be made open source so that anyone who feels like it could investigate it.

Who would have access to the “sensitive” data like IP addresses of the servers, i.e. the data which would not be publicly disclosed? Given recent revelations in the news, how would possible data requests from third parties be handled (such as a list of IP addresses for where DHIS2 is used)?

We will simply not store IP addresses. We are not interested in them anyway, and to avoid any thinkable doubt of misuse or NSA inquiries we will not store them.

I am thinking that for this function to have a purpose, it should not be possible to opt out of the very basic, anonymous data, like DHIS version and random system identifier. If optional then the statistics wouldn’t be very useful. If that is not acceptable we would rather drop the feature. Of course, the “activation” feature with contact person etc would be completely optional and require that you actively enable it.

One way to approach IP address security is to not record them anywhere (and make sure the central server to which the data is sent does not keep any log of them.) If the central server doesn’t know the IP addresses, they can’t be divulged to any third party.

Agreed.

A very different approach to openness and security is taken by the optional OpenMRS “Atlas module”. This opt-in module allows an OpenMRS installation to provide information that is available in a public OpenMRS “atlas” of implementations. The atlas can facilitate interactions between OpenMRS community members, including prospective members, and it also serves as publicity for OpenMRS as well as the implementers. I’ll leave it for others to say whether DHIS2 administrators would feel comfortable or even welcome this kind of public information for their implementations. See http://openmrs.org/atlas/ for the OpenMRS atlas itself, and https://wiki.openmrs.org/display/docs/Atlas+Module+User+Guide for the atlas module user guide.

Thanks for the tip, I will check out the atlas module for inspiration.

Another idea for this concept: The DHIS instances could be checking with a central system whether new major versions of DHIS are available for download or whether any critical bugfixes are available for the current major version. This notifications could be made available through the DHIS messaging system and be sent to some group of system admin users.

More comments are welcome.

cheers

Lars