You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@impala.apache.org by Bharath Vissapragada <bh...@cloudera.com> on 2016/09/06 01:27:42 UTC
Jira move to apache - Inputs required
Hello folks and mentors,
As a part of Impala's transition to ASF I'm working
<https://issues.cloudera.org/browse/IMPALA-3224> on moving the existing
jira infra from cloudera.org to issues.apache.org. As a part of this move,
we need to map the content from existing jira instance to the one in
apache. This requires mapping the user names between these instances.
Recently I've sent out a form
<https://docs.google.com/forms/d/e/1FAIpQLSe_pa17Swc1y4nJZvoFJ3Wek10hFbxPenWnWDkBvHS-6P1Cww/viewform?c=0&w=1>
for the users to submit this mapping. I only received responses from ~30
people so far.
Running some analysis on the current jira dump, we have the following stats.
(1) A total of ~750 users
(2) Out of these ~750 users, ~40 users have no kind of activity in the
jira. This activity includes creating/assignees/watching/commenting on the
jiras.
(3) Out of the remaining ~750 users, ~250 people are just watchers for
jiras and do not have any other kind of activity.
Excluding just watchers, we have ~500 registered users with any kind of
real activity on the jira. Excluding the ~30 users who filled the above
form, we have ~470 people with some kind of activity which will lost in the
transition if no user mapping is provided. Unfortunately this list is too
big to list to do any kind of manual verification and adding the entries
myself after verifying their apache ids.
Do you have any inputs on what can be done with these un-mapped users? We
have the following options
(a) Create a dummy user and attribute all the content from these unmapped
users activity to it.
(b) Create a new jira id for these unmapped users. Lets stay if the user id
is john, we create john_impala (or something similar) by appending a random
string so that we don't map the content incorrectly to a user with similar
name on the apache jira.
I'm inclined towards (b) as it saves the user activity somewhere and can be
recovered by the users by providing the email id and resetting the password
if they are really interested in it. However it is unclear if they can
export it and import into their existing apache account if required and
hence not really sure if it will be of any use.
Mentors or committers or fellow devs, do you have any preferences on what
can be done in this case?
Re: Jira move to apache - Inputs required
Posted by Mike Percy <mp...@cloudera.com>.
I chatted with Bharath offline last week and gave him all the details (and
scripts) that we used for the Kudu JIRA migration, as well as my own ideas
about how this could be done. I'm happy to help further, just reach out if
you need anything.
Mike
--
Mike Percy
Software Engineer, Cloudera
On Tue, Sep 6, 2016 at 10:43 AM, Bharath Vissapragada <bharathv@cloudera.com
> wrote:
> Thanks Brock and Lars.
>
> @Lars I had a chat with Mike (cc'ed) the other day and looks like this
> unmapped list was quite small in Kudu's case. IIRC Mike verified the users
> manually to some extent by visiting the apache websites corresponding to
> the users. Not sure if it is feasible in Impala's case though.
>
> On Tue, Sep 6, 2016 at 3:54 AM, Lars Volker <lv...@cloudera.com> wrote:
>
>> Quoting a mail from Mike on the subject a while back. I will forward you
>> the user mappings separately.
>>
>> TLDR; Kudu used b).
>>
>> Hi Lars, please see inline.
>> > On Fri, Jun 24, 2016 at 12:16 PM, Lars Volker <lv...@cloudera.com> wrote:
>> > Taras and me are working on migrating the Impala Jira instance to the
>> ASF
>> > Jira IMPALA-3224 <https://issues.cloudera.org/browse/IMPALA-3224>. I
>> had
>> > a look at the scripts you used to do the same for Kudu and have a few
>> > questions that you could maybe help us with:
>> >
>> > - You mentioned in the Jira that we could use your mapping files as a
>> > starting point. Could you share them with us?
>> >
>> > Forwarded to you via email.
>> >
>> > - We probably have external users in our Jira who created an account
>> > at some point in time, opened an issue and disappeared for good. Now
>> we are
>> > reluctant to create users for those on the ASF Jira, not wanting to "
>> > impersonate" them. How did you handle users you were not able to get
>> > ahold of?
>> >
>> > I sent out a lot of emails and waited about a week. I also searched
>> around
>> > trying to figure out what their accounts were. The people who didn't
>> > respond fell into two categories:
>> >
>> > 1. They had the same username as someone else in the ASF JIRA
>> instance
>> >
>> >
>> > 1. They had an "unclaimed" username in the ASF JIRA instance
>>
>> >
>> > For #1, luckily I didn't have to deal with this because people responded
>> > to my emails eventually, but I would have just picked a new username for
>> > them and put it in the mapping file. Then they become #2...
>> > For #2, it's not a very big deal to create an account for them. They
>> could
>> > recover their password and regain access to their account later if they
>> > have access to the email address they had used to register. If they
>> don't,
>> > and they want to claim the account, they can just file an ASF INFRA
>> ticket
>> > and the ASF admins can work it out with them.
>> >
>>
>> On Tue, Sep 6, 2016 at 3:50 AM, Brock Noland <br...@phdata.io> wrote:
>>
>> > Flume used a, see here:
>> >
>> > https://issues.apache.org/jira/browse/FLUME-3
>> >
>> > On Mon, Sep 5, 2016 at 8:27 PM, Bharath Vissapragada
>> > <bh...@cloudera.com> wrote:
>> > > Hello folks and mentors,
>> > >
>> > > As a part of Impala's transition to ASF I'm working
>> > > <https://issues.cloudera.org/browse/IMPALA-3224> on moving the
>> existing
>> > > jira infra from cloudera.org to issues.apache.org. As a part of this
>> > move,
>> > > we need to map the content from existing jira instance to the one in
>> > > apache. This requires mapping the user names between these instances.
>> > > Recently I've sent out a form
>> > > <https://docs.google.com/forms/d/e/1FAIpQLSe_
>> > pa17Swc1y4nJZvoFJ3Wek10hFbxPenWnWDkBvHS-6P1Cww/viewform?c=0&w=1>
>> > > for the users to submit this mapping. I only received responses from
>> ~30
>> > > people so far.
>> > >
>> > > Running some analysis on the current jira dump, we have the following
>> > stats.
>> > >
>> > > (1) A total of ~750 users
>> > > (2) Out of these ~750 users, ~40 users have no kind of activity in the
>> > > jira. This activity includes creating/assignees/watching/commenting
>> on
>> > the
>> > > jiras.
>> > > (3) Out of the remaining ~750 users, ~250 people are just watchers for
>> > > jiras and do not have any other kind of activity.
>> > >
>> > > Excluding just watchers, we have ~500 registered users with any kind
>> of
>> > > real activity on the jira. Excluding the ~30 users who filled the
>> above
>> > > form, we have ~470 people with some kind of activity which will lost
>> in
>> > the
>> > > transition if no user mapping is provided. Unfortunately this list is
>> > too
>> > > big to list to do any kind of manual verification and adding the
>> entries
>> > > myself after verifying their apache ids.
>> > >
>> > > Do you have any inputs on what can be done with these un-mapped
>> users? We
>> > > have the following options
>> > >
>> > > (a) Create a dummy user and attribute all the content from these
>> > unmapped
>> > > users activity to it.
>> > >
>> > > (b) Create a new jira id for these unmapped users. Lets stay if the
>> user
>> > id
>> > > is john, we create john_impala (or something similar) by appending a
>> > random
>> > > string so that we don't map the content incorrectly to a user with
>> > similar
>> > > name on the apache jira.
>> > >
>> > > I'm inclined towards (b) as it saves the user activity somewhere and
>> can
>> > be
>> > > recovered by the users by providing the email id and resetting the
>> > password
>> > > if they are really interested in it. However it is unclear if they can
>> > > export it and import into their existing apache account if required
>> and
>> > > hence not really sure if it will be of any use.
>> > >
>> > > Mentors or committers or fellow devs, do you have any preferences on
>> what
>> > > can be done in this case?
>> >
>>
>
>
Re: Jira move to apache - Inputs required
Posted by Bharath Vissapragada <bh...@cloudera.com>.
Thanks Brock and Lars.
@Lars I had a chat with Mike (cc'ed) the other day and looks like this
unmapped list was quite small in Kudu's case. IIRC Mike verified the users
manually to some extent by visiting the apache websites corresponding to
the users. Not sure if it is feasible in Impala's case though.
On Tue, Sep 6, 2016 at 3:54 AM, Lars Volker <lv...@cloudera.com> wrote:
> Quoting a mail from Mike on the subject a while back. I will forward you
> the user mappings separately.
>
> TLDR; Kudu used b).
>
> Hi Lars, please see inline.
> > On Fri, Jun 24, 2016 at 12:16 PM, Lars Volker <lv...@cloudera.com> wrote:
> > Taras and me are working on migrating the Impala Jira instance to the ASF
> > Jira IMPALA-3224 <https://issues.cloudera.org/browse/IMPALA-3224>. I had
> > a look at the scripts you used to do the same for Kudu and have a few
> > questions that you could maybe help us with:
> >
> > - You mentioned in the Jira that we could use your mapping files as a
> > starting point. Could you share them with us?
> >
> > Forwarded to you via email.
> >
> > - We probably have external users in our Jira who created an account
> > at some point in time, opened an issue and disappeared for good. Now
> we are
> > reluctant to create users for those on the ASF Jira, not wanting to "
> > impersonate" them. How did you handle users you were not able to get
> > ahold of?
> >
> > I sent out a lot of emails and waited about a week. I also searched
> around
> > trying to figure out what their accounts were. The people who didn't
> > respond fell into two categories:
> >
> > 1. They had the same username as someone else in the ASF JIRA instance
> >
> >
> > 1. They had an "unclaimed" username in the ASF JIRA instance
> >
> > For #1, luckily I didn't have to deal with this because people responded
> > to my emails eventually, but I would have just picked a new username for
> > them and put it in the mapping file. Then they become #2...
> > For #2, it's not a very big deal to create an account for them. They
> could
> > recover their password and regain access to their account later if they
> > have access to the email address they had used to register. If they
> don't,
> > and they want to claim the account, they can just file an ASF INFRA
> ticket
> > and the ASF admins can work it out with them.
> >
>
> On Tue, Sep 6, 2016 at 3:50 AM, Brock Noland <br...@phdata.io> wrote:
>
> > Flume used a, see here:
> >
> > https://issues.apache.org/jira/browse/FLUME-3
> >
> > On Mon, Sep 5, 2016 at 8:27 PM, Bharath Vissapragada
> > <bh...@cloudera.com> wrote:
> > > Hello folks and mentors,
> > >
> > > As a part of Impala's transition to ASF I'm working
> > > <https://issues.cloudera.org/browse/IMPALA-3224> on moving the
> existing
> > > jira infra from cloudera.org to issues.apache.org. As a part of this
> > move,
> > > we need to map the content from existing jira instance to the one in
> > > apache. This requires mapping the user names between these instances.
> > > Recently I've sent out a form
> > > <https://docs.google.com/forms/d/e/1FAIpQLSe_
> > pa17Swc1y4nJZvoFJ3Wek10hFbxPenWnWDkBvHS-6P1Cww/viewform?c=0&w=1>
> > > for the users to submit this mapping. I only received responses from
> ~30
> > > people so far.
> > >
> > > Running some analysis on the current jira dump, we have the following
> > stats.
> > >
> > > (1) A total of ~750 users
> > > (2) Out of these ~750 users, ~40 users have no kind of activity in the
> > > jira. This activity includes creating/assignees/watching/commenting on
> > the
> > > jiras.
> > > (3) Out of the remaining ~750 users, ~250 people are just watchers for
> > > jiras and do not have any other kind of activity.
> > >
> > > Excluding just watchers, we have ~500 registered users with any kind of
> > > real activity on the jira. Excluding the ~30 users who filled the above
> > > form, we have ~470 people with some kind of activity which will lost in
> > the
> > > transition if no user mapping is provided. Unfortunately this list is
> > too
> > > big to list to do any kind of manual verification and adding the
> entries
> > > myself after verifying their apache ids.
> > >
> > > Do you have any inputs on what can be done with these un-mapped users?
> We
> > > have the following options
> > >
> > > (a) Create a dummy user and attribute all the content from these
> > unmapped
> > > users activity to it.
> > >
> > > (b) Create a new jira id for these unmapped users. Lets stay if the
> user
> > id
> > > is john, we create john_impala (or something similar) by appending a
> > random
> > > string so that we don't map the content incorrectly to a user with
> > similar
> > > name on the apache jira.
> > >
> > > I'm inclined towards (b) as it saves the user activity somewhere and
> can
> > be
> > > recovered by the users by providing the email id and resetting the
> > password
> > > if they are really interested in it. However it is unclear if they can
> > > export it and import into their existing apache account if required and
> > > hence not really sure if it will be of any use.
> > >
> > > Mentors or committers or fellow devs, do you have any preferences on
> what
> > > can be done in this case?
> >
>
Re: Jira move to apache - Inputs required
Posted by Lars Volker <lv...@cloudera.com>.
Quoting a mail from Mike on the subject a while back. I will forward you
the user mappings separately.
TLDR; Kudu used b).
Hi Lars, please see inline.
> On Fri, Jun 24, 2016 at 12:16 PM, Lars Volker <lv...@cloudera.com> wrote:
> Taras and me are working on migrating the Impala Jira instance to the ASF
> Jira IMPALA-3224 <https://issues.cloudera.org/browse/IMPALA-3224>. I had
> a look at the scripts you used to do the same for Kudu and have a few
> questions that you could maybe help us with:
>
> - You mentioned in the Jira that we could use your mapping files as a
> starting point. Could you share them with us?
>
> Forwarded to you via email.
>
> - We probably have external users in our Jira who created an account
> at some point in time, opened an issue and disappeared for good. Now we are
> reluctant to create users for those on the ASF Jira, not wanting to "
> impersonate" them. How did you handle users you were not able to get
> ahold of?
>
> I sent out a lot of emails and waited about a week. I also searched around
> trying to figure out what their accounts were. The people who didn't
> respond fell into two categories:
>
> 1. They had the same username as someone else in the ASF JIRA instance
>
>
> 1. They had an "unclaimed" username in the ASF JIRA instance
>
> For #1, luckily I didn't have to deal with this because people responded
> to my emails eventually, but I would have just picked a new username for
> them and put it in the mapping file. Then they become #2...
> For #2, it's not a very big deal to create an account for them. They could
> recover their password and regain access to their account later if they
> have access to the email address they had used to register. If they don't,
> and they want to claim the account, they can just file an ASF INFRA ticket
> and the ASF admins can work it out with them.
>
On Tue, Sep 6, 2016 at 3:50 AM, Brock Noland <br...@phdata.io> wrote:
> Flume used a, see here:
>
> https://issues.apache.org/jira/browse/FLUME-3
>
> On Mon, Sep 5, 2016 at 8:27 PM, Bharath Vissapragada
> <bh...@cloudera.com> wrote:
> > Hello folks and mentors,
> >
> > As a part of Impala's transition to ASF I'm working
> > <https://issues.cloudera.org/browse/IMPALA-3224> on moving the existing
> > jira infra from cloudera.org to issues.apache.org. As a part of this
> move,
> > we need to map the content from existing jira instance to the one in
> > apache. This requires mapping the user names between these instances.
> > Recently I've sent out a form
> > <https://docs.google.com/forms/d/e/1FAIpQLSe_
> pa17Swc1y4nJZvoFJ3Wek10hFbxPenWnWDkBvHS-6P1Cww/viewform?c=0&w=1>
> > for the users to submit this mapping. I only received responses from ~30
> > people so far.
> >
> > Running some analysis on the current jira dump, we have the following
> stats.
> >
> > (1) A total of ~750 users
> > (2) Out of these ~750 users, ~40 users have no kind of activity in the
> > jira. This activity includes creating/assignees/watching/commenting on
> the
> > jiras.
> > (3) Out of the remaining ~750 users, ~250 people are just watchers for
> > jiras and do not have any other kind of activity.
> >
> > Excluding just watchers, we have ~500 registered users with any kind of
> > real activity on the jira. Excluding the ~30 users who filled the above
> > form, we have ~470 people with some kind of activity which will lost in
> the
> > transition if no user mapping is provided. Unfortunately this list is
> too
> > big to list to do any kind of manual verification and adding the entries
> > myself after verifying their apache ids.
> >
> > Do you have any inputs on what can be done with these un-mapped users? We
> > have the following options
> >
> > (a) Create a dummy user and attribute all the content from these
> unmapped
> > users activity to it.
> >
> > (b) Create a new jira id for these unmapped users. Lets stay if the user
> id
> > is john, we create john_impala (or something similar) by appending a
> random
> > string so that we don't map the content incorrectly to a user with
> similar
> > name on the apache jira.
> >
> > I'm inclined towards (b) as it saves the user activity somewhere and can
> be
> > recovered by the users by providing the email id and resetting the
> password
> > if they are really interested in it. However it is unclear if they can
> > export it and import into their existing apache account if required and
> > hence not really sure if it will be of any use.
> >
> > Mentors or committers or fellow devs, do you have any preferences on what
> > can be done in this case?
>
Re: Jira move to apache - Inputs required
Posted by Brock Noland <br...@phdata.io>.
Flume used a, see here:
https://issues.apache.org/jira/browse/FLUME-3
On Mon, Sep 5, 2016 at 8:27 PM, Bharath Vissapragada
<bh...@cloudera.com> wrote:
> Hello folks and mentors,
>
> As a part of Impala's transition to ASF I'm working
> <https://issues.cloudera.org/browse/IMPALA-3224> on moving the existing
> jira infra from cloudera.org to issues.apache.org. As a part of this move,
> we need to map the content from existing jira instance to the one in
> apache. This requires mapping the user names between these instances.
> Recently I've sent out a form
> <https://docs.google.com/forms/d/e/1FAIpQLSe_pa17Swc1y4nJZvoFJ3Wek10hFbxPenWnWDkBvHS-6P1Cww/viewform?c=0&w=1>
> for the users to submit this mapping. I only received responses from ~30
> people so far.
>
> Running some analysis on the current jira dump, we have the following stats.
>
> (1) A total of ~750 users
> (2) Out of these ~750 users, ~40 users have no kind of activity in the
> jira. This activity includes creating/assignees/watching/commenting on the
> jiras.
> (3) Out of the remaining ~750 users, ~250 people are just watchers for
> jiras and do not have any other kind of activity.
>
> Excluding just watchers, we have ~500 registered users with any kind of
> real activity on the jira. Excluding the ~30 users who filled the above
> form, we have ~470 people with some kind of activity which will lost in the
> transition if no user mapping is provided. Unfortunately this list is too
> big to list to do any kind of manual verification and adding the entries
> myself after verifying their apache ids.
>
> Do you have any inputs on what can be done with these un-mapped users? We
> have the following options
>
> (a) Create a dummy user and attribute all the content from these unmapped
> users activity to it.
>
> (b) Create a new jira id for these unmapped users. Lets stay if the user id
> is john, we create john_impala (or something similar) by appending a random
> string so that we don't map the content incorrectly to a user with similar
> name on the apache jira.
>
> I'm inclined towards (b) as it saves the user activity somewhere and can be
> recovered by the users by providing the email id and resetting the password
> if they are really interested in it. However it is unclear if they can
> export it and import into their existing apache account if required and
> hence not really sure if it will be of any use.
>
> Mentors or committers or fellow devs, do you have any preferences on what
> can be done in this case?