Ori is in the process of obsoleting the current modality of the ClickTracking extension and replaced at some point with what the E3 team is building for collection on Vanadium (and then folded into Kraken as that comes available).
Depending on ops and the handling of the data collection piece, this may take a while, but I am asking if anyone is actually using ClickTracking for anything anymore? I know some things are/may still be recording (AFTv5, MoodBar, etc.), but to my knowledge none of this data is supposed to be being used at this point. This does not cover the clicktracking that is currently being done on the E3 team (obviously), but does cover past editor engagement stuff.
If that is not the case, then reply to this thread with what and how long and Ori will make sure that collection for those will be handled somehow for the duration it is needed or some sort of replacement implementation is done. Otherwise, let's let those streams die.
Note this only covers data generated by the clicktracking extension specifically.
AFTv5 is using ClickTracking according to this page: http://www.mediawiki.org/wiki/Extension:ArticleFeedbackv5
-Chris
On Fri, Jul 27, 2012 at 2:37 PM, Terry Chay tchay@wikimedia.org wrote:
Ori is in the process of obsoleting the current modality of theClickTracking extension and replaced at some point with what the E3 team is building for collection on Vanadium (and then folded into Kraken as that comes available).
Depending on ops and the handling of the data collection piece,this may take a while, but I am asking if anyone is actually using ClickTracking for anything anymore? I know some things are/may still be recording (AFTv5, MoodBar, etc.), but to my knowledge none of this data is supposed to be being used at this point. This does not cover the clicktracking that is currently being done on the E3 team (obviously), but does cover past editor engagement stuff.
If that is not the case, then reply to this thread with what andhow long and Ori will make sure that collection for those will be handled somehow for the duration it is needed or some sort of replacement implementation is done. Otherwise, let's let those streams die.
Note this only covers data generated by the clicktrackingextension specifically. _______________________________________________ ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
MoodBar is using clickTracking to track the clicks on the feedback response email
On Fri, Jul 27, 2012 at 1:37 PM, Terry Chay tchay@wikimedia.org wrote:
Ori is in the process of obsoleting the current modality of theClickTracking extension and replaced at some point with what the E3 team is building for collection on Vanadium (and then folded into Kraken as that comes available).
Depending on ops and the handling of the data collection piece,this may take a while, but I am asking if anyone is actually using ClickTracking for anything anymore? I know some things are/may still be recording (AFTv5, MoodBar, etc.), but to my knowledge none of this data is supposed to be being used at this point. This does not cover the clicktracking that is currently being done on the E3 team (obviously), but does cover past editor engagement stuff.
If that is not the case, then reply to this thread with what andhow long and Ori will make sure that collection for those will be handled somehow for the duration it is needed or some sort of replacement implementation is done. Otherwise, let's let those streams die.
Note this only covers data generated by the clicktrackingextension specifically. _______________________________________________ ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
We removed all clicktracking from AFT5 and do not have it for NewPagesFeed either.
When Dario gets back to work on Monday, he can let us know how much longer it needs to stay on Moodbar.
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
Cheers,
Fabrice
On Jul 27, 2012, at 1:58 PM, Benny Situ wrote:
MoodBar is using clickTracking to track the clicks on the feedback response email
On Fri, Jul 27, 2012 at 1:37 PM, Terry Chay tchay@wikimedia.org wrote: Ori is in the process of obsoleting the current modality of the ClickTracking extension and replaced at some point with what the E3 team is building for collection on Vanadium (and then folded into Kraken as that comes available).
Depending on ops and the handling of the data collection piece, this may take a while, but I am asking if anyone is actually using ClickTracking for anything anymore? I know some things are/may still be recording (AFTv5, MoodBar, etc.), but to my knowledge none of this data is supposed to be being used at this point. This does not cover the clicktracking that is currently being done on the E3 team (obviously), but does cover past editor engagement stuff. If that is not the case, then reply to this thread with what and how long and Ori will make sure that collection for those will be handled somehow for the duration it is needed or some sort of replacement implementation is done. Otherwise, let's let those streams die. Note this only covers data generated by the clicktracking extension specifically.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
On Fri, Jul 27, 2012 at 3:24 PM, Fabrice Florin fflorin@wikimedia.orgwrote:
We removed all clicktracking from AFT5 and do not have it for NewPagesFeed either.
I'm not entirely certain of this because I'm still exploring getting AFTv5 up and running on the beta cluster, but it may be the case that ClickTracking (and its dependency, UserDailyContribs), may have to be in place, but disabled, for AFTv5. Again, I'm not 100% certain of this, but that is the current setup right now.
-Chris
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis. ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call). ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15. E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
also note that logging has simply been disabled but bits of code calling the clicktracking API are lurking everywhere: if we want to discontinue it entirely we will need to clean up lots of extensions that we have in production, on top of those listed below. For example WikiEditor still has the ability of tracking button clicks.
On Jul 27, 2012, at 3:11 PM, Dario Taraborelli wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis. ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call). ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15. E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
Couldn't we just have E3_Experiments have those functions and do a no-op? If they're hooks then they can safely be ignored. :-)
On Jul 27, 2012, at 3:25 PM, Dario Taraborelli dtaraborelli@wikimedia.org wrote:
also note that logging has simply been disabled but bits of code calling the clicktracking API are lurking everywhere: if we want to discontinue it entirely we will need to clean up lots of extensions that we have in production, on top of those listed below. For example WikiEditor still has the ability of tracking button clicks.
On Jul 27, 2012, at 3:11 PM, Dario Taraborelli wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis. ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call). ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15. E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
Event IDs logged using ClickTracking extension in the past 30 days: http://dpaste.com/hold/776542/
Dario, I think your list accounts for all of them, except perhaps the ext.vector.* and leftnav-vector-* event types. Do you know what those are?
On Friday, July 27, 2012 at 3:11 PM, Dario Taraborelli wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin <fflorin@wikimedia.org (mailto:fflorin@wikimedia.org)> wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org (mailto:ee@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/ee
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics
Ori, I think the Vector extension is creating those events.
On Fri, Jul 27, 2012 at 3:55 PM, Ori Livneh ori.livneh@gmail.com wrote:
Event IDs logged using ClickTracking extension in the past 30 days: http://dpaste.com/hold/776542/
Dario, I think your list accounts for all of them, except perhaps the ext.vector.* and leftnav-vector-* event types. Do you know what those are?
On Friday, July 27, 2012 at 3:11 PM, Dario Taraborelli wrote:
Thanks for starting this thread. Diederik and I already shut down
several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into
emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki;
enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users
running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
ext.MoodBar collecting email notification click-through data from enwiki, will be
disabled at the latest by August 15.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be
moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the
ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and
analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this
can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if
other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin <fflorin@wikimedia.org(mailto:
fflorin@wikimedia.org)> wrote:
However, we will want to start collecting clicktracking data again
soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org (mailto:ee@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/ee
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
ext.vector.sectionEditLinks: that's Trevor's old SectionEditLink experiment, but from what I see these events come from testwiki only
leftnav-vector-*: no clue, but isn't it pointless to try and reconstruct the source of events that do not correctly identify themselves? I just logged an event "ilovespaghetti" with a click on this url: http://en.wikipedia.org/w/api.php?action=clicktracking&eventid=ilovespag...
dartar@emery:/var/log/aft$ grep ilovespaghetti clicktracking.log enwiki ilovespaghetti 20120727232404 1 klp6Dw8dqhHQrVVYJUkqF3mvLpUUlF2yf 0 2589 347 179 59
Dario
On Jul 27, 2012, at 3:55 PM, Ori Livneh wrote:
Event IDs logged using ClickTracking extension in the past 30 days: http://dpaste.com/hold/776542/
Dario, I think your list accounts for all of them, except perhaps the ext.vector.* and leftnav-vector-* event types. Do you know what those are?
On Friday, July 27, 2012 at 3:11 PM, Dario Taraborelli wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin <fflorin@wikimedia.org (mailto:fflorin@wikimedia.org)> wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org (mailto:ee@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/ee
Analytics mailing list Analytics@lists.wikimedia.org (mailto:Analytics@lists.wikimedia.org) https://lists.wikimedia.org/mailman/listinfo/analytics
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics
On Jul 27, 2012, at 3:11 PM, Dario Taraborelli dtaraborelli@wikimedia.org wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
Yes, let's let this one die and see if anyone complains. BTW, orthogonal to this discussion we should see is the possibility of turning off AFTv4 entirely. The reason is AFTv5 to 100% will be blocked on performance considerations, but I got the impression that English doesn't even care for AFTv4 anymore.
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
Whee! Let's plan that if Fabrice needs something new, they will be handled by the E3_experiments extension going forward.
Do we want a cross functional presentation or sprint on this E3_experiments or Kraken on the Tech Day in September? I noticed that nothing analytics is in the parking lot yet https://www.mediawiki.org/wiki/Wikimedia_Engineering/WMF_Tech_Days_2012#Wishlist_.2F_parking_lot.
ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15.
Ori can decide if this needs to be moved.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
On Jul 30, 2012, at 2:33 PM, Terry Chay wrote:
On Jul 27, 2012, at 3:11 PM, Dario Taraborelli dtaraborelli@wikimedia.org wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
Yes, let's let this one die and see if anyone complains. BTW, orthogonal to this discussion we should see is the possibility of turning off AFTv4 entirely. The reason is AFTv5 to 100% will be blocked on performance considerations, but I got the impression that English doesn't even care for AFTv4 anymore.
Yes, my understanding is that we are ready to disable AFT4 entirely, at least from a WMF product team standpoint.
I announced in two successive blog posts that we would retire it in September, when we deploy AFT5 to 100%, so people who read our blog may already be aware of this planned change.
But we may want to do one final community outreach before turning it off completely.
Does anyone have recommendations on how we might do this final communication? I suspect the folks who would care the most may be researchers, rather than community members ...
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
Whee! Let's plan that if Fabrice needs something new, they will be handled by the E3_experiments extension going forward.
Sounds good to me.
When do we expect to have access to the E3 clicktracking function? We may want to collect some data on the feedback page after we deploy successfully to 100%, which could be anywhere from mid-September to late October, depending on how our scaleability analysis goes in a couple weeks.
Do we want a cross functional presentation or sprint on this E3_experiments or Kraken on the Tech Day in September? I noticed that nothing analytics is in the parking lot yet https://www.mediawiki.org/wiki/Wikimedia_Engineering/WMF_Tech_Days_2012#Wishlist_.2F_parking_lot.
ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15.
Ori can decide if this needs to be moved.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
let me clarify that when we say: "we are ready to disable AFT4 entirely, at least from a WMF product team standpoint" this only applies to enwiki. AFAIK we have no strategy or clear ownership in place for non-enwiki editor engagement features.
Regarding what to do with the data once we discontinue AFT4 I will release on the datahub a static dataset for research purposes when AFT5 eventually takes over.
Dario
On Jul 30, 2012, at 3:16 PM, Fabrice Florin wrote:
On Jul 30, 2012, at 2:33 PM, Terry Chay wrote:
On Jul 27, 2012, at 3:11 PM, Dario Taraborelli dtaraborelli@wikimedia.org wrote:
Thanks for starting this thread. Diederik and I already shut down several undocumented data collection jobs that nobody owned months ago (one of the problems of CT is that anything/anyone can start dumping data into the log by calling the API calls, without identifying itself).
These are features currently logging events into emery:/var/aft/log/clicktracking.log
ext.articleFeedback (AFT4) currently collecting data from enwiki, eswiki, ptwiki, zhwiki, testwiki; enwiki will be entirely disabled with the completion of the AFTv5 ramp-up, I don't know why we are still collecting data from other wikis unless this was commissioned by Global Dev. I recommend we discontinue AFTv4 logging on all wikis.
Yes, let's let this one die and see if anyone complains. BTW, orthogonal to this discussion we should see is the possibility of turning off AFTv4 entirely. The reason is AFTv5 to 100% will be blocked on performance considerations, but I got the impression that English doesn't even care for AFTv4 anymore.
Yes, my understanding is that we are ready to disable AFT4 entirely, at least from a WMF product team standpoint.
I announced in two successive blog posts that we would retire it in September, when we deploy AFT5 to 100%, so people who read our blog may already be aware of this planned change.
But we may want to do one final community outreach before turning it off completely.
Does anyone have recommendations on how we might do this final communication? I suspect the folks who would care the most may be researchers, rather than community members ...
ext.articleFeedbackv5 (AFT5) entirely disabled on enwiki, but events keep trickling in from users running stale code. It can be safely discontinued unless there's a need of resuming data collection for FeedbackPage usage or CTAs after the completion of the ramp-up (Fabrice's call).
Whee! Let's plan that if Fabrice needs something new, they will be handled by the E3_experiments extension going forward.
Sounds good to me.
When do we expect to have access to the E3 clicktracking function? We may want to collect some data on the feedback page after we deploy successfully to 100%, which could be anywhere from mid-September to late October, depending on how our scaleability analysis goes in a couple weeks.
Do we want a cross functional presentation or sprint on this E3_experiments or Kraken on the Tech Day in September? I noticed that nothing analytics is in the parking lot yet https://www.mediawiki.org/wiki/Wikimedia_Engineering/WMF_Tech_Days_2012#Wishlist_.2F_parking_lot.
ext.MoodBar collecting email notification click-through data from enwiki, will be disabled at the latest by August 15.
Ori can decide if this needs to be moved.
E3 experiments (PEF, Community Portal) collecting clickthrough and edit save completion data on enwiki, will be moved to vanadium as soon as it's ready to collect data.
I am not aware of any other features/gadgets currently using the ClickTracking extension.
Dario
On Jul 27, 2012, at 2:42 PM, Terry Chay wrote:
The idea is to fold all of editor engagement data collection and analysis into the tool/services that Ori is building out on vanadium. If we can shut down clicktracking and udp2log, then we can not only start collecting this data again soon, but also increase what we collect, and do analytics faster.
In the short term, since this is editor data (logged in usage), this can all be handled on vanadium. In the longer term, these will be queries on Kraken.
BTW, I don't know when it would be "removed" but at the minimum, if other things are delayed, clicktracking might get "broken" for old data streams unless we specifically support them.
On Jul 27, 2012, at 2:24 PM, Fabrice Florin fflorin@wikimedia.org wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
ee mailing list ee@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/ee
On Friday, July 27, 2012 at 2:24 PM, Fabrice Florin wrote:
However, we will want to start collecting clicktracking data again soon, so my hope is that we can have a solution for that in coming weeks.
Provided the volume of events is comparable to current / historic levels, I can commit to making sure something is available.