updated scoreboard of dallas regional

[ronlottem]

do anyone have an updated scoreboard of dallas?
the blue alliance one only puts question marks like the matches didnt start yet.

[plnyyanks]

do anyone have an updated scoreboard of dallas?
the blue alliance one only puts question marks like the matches didnt start yet.

You're going to have to ask somebody who was there - the FRC API has been broken all day, nobody has been able to get any match info at all, making TBA, Spyder, etc all without data.

You can check this results site (http://frc-events.usfirst.org/), some events have been able to get rankings up throughout the day. The may or may not be up to date, though.

Hopefully all of these data issues will be resolved soon

[MooreteP]

Welcome to week one.

[twetherbee]

Rankings read off by the announcer at the end of the day:

1) 987 101.5 avg
2) 148 81.4 avg
3) 118 77.2 avg
4) 1477 76 avg
5) 4300 (only gave the top 4 averages)
6) 4499
7) 2950
8) 647
9) 704
10) 457
11) 3350
12) 5431

[Zach101]

The FMS that sends info to FRC spyder and TBA isn't working. At least for northern lights and lake Superior regionals. Nothing has been updated in at least 1 and a half hours. Probably more.

[Boltman]

Rankings read off by the announcer at the end of the day:

1) 987 101.5 avg
2) 148 81.4 avg
3) 118 77.2 avg
4) 1477 76 avg
5) 4300 (only gave the top 4 averages)
6) 4499
7) 2950
8) 647
9) 704
10) 457
11) 3350
12) 5431

Thats much lower than I expected , I thought for sure Dallas would be at 120 for the top alliance. Disappointing.
Oh well see what overnight adjustments do. Congrats Las Vegas!

[Mike Soukup]

Welcome to week one.

It's sad that this has been a common theme for so many years. I guess this is what happens when you only get your server online a few days before events start. That's not enough time to load test it or fix the inevitable bugs.

[asid61]

Definitely disappointed. I hope it picks up tomorrow (it likely will at least a bit IME).
Rankings are here: http://frc-events.usfirst.org/2015/TXDA/rankings
However, it's nice that the gap between really good and just good teams isn't incredibly high. I want to see some video of those matches to see how the actual event is working out.

[plnyyanks]

It's sad that this has been a common theme for so many years. I guess this is what happens when you only get your server online a few days before events start. That's not enough time to load test it or fix the inevitable bugs.

What's even more frustrating is the fact that there has been almost zero communication all day about the problems, or an estimated fix timeframe. All we've heard is a post of TeamForge saying they're working on it at 10 AM. I get they're busy with competitions and all, but a "hey, our database melted, but it should be back up tomorrow" isn't that time consuming.

Hopefully they get everything ironed out quickly; I think I'd prefer the old HTML pages to a buggy API.

[wesbass23]

Thats much lower than I expected , I thought for sure Dallas would be at 120 for the top alliance. Disappointing.
Oh well see what overnight adjustments do. Congrats Las Vegas!

In multiple matches even the top teams listed at Dallas did not go for coop pts (in some cases they did try for them but the other alliance did not contribute) which could lift those average scores by a fair amount.

[Zach101]

In multiple matches even the top teams listed at Dallas did not go for coop pts (in some cases they did try for them but the other alliance did not contribute) which could lift those average scores by a fair amount.

Co-op points are huge for the northern lights regional. Most(a lot) of qualification matches there would be a robot from each alliance going for co-op.

[Mike Soukup]

What's even more frustrating is the fact that there has been almost zero communication all day about the problems, or an estimated fix timeframe. All we've heard is a post of TeamForge saying they're working on it at 10 AM. I get they're busy with competitions and all, but a "hey, our database melted, but it should be back up tomorrow" isn't that time consuming.

Hopefully they get everything ironed out quickly; I think I'd prefer the old HTML pages to a buggy API.

My fear is that no news is bad news. If they knew how to fix the problem and how long it would take, they'd probably let us know on TeamForge.

Maybe they can just push the results to Spyder or TBA directly, we both know how to handle the load ;)

[plnyyanks]

My fear is that no news is bad news. If they knew how to fix the problem and how long it would take, they'd probably let us know on TeamForge.

Maybe they can just push the results to Spyder or TBA directly, we both know how to handle the load ;)

That's probably the unfortunate truth. My guess is that caching was lacking and couldn't handle all the load - some of the endpoints get really specific with lots of parameters. I can see that being difficult to cache properly, especially when the data is updating rapidly.

We should also get Chezy Arena to replace FMS and get a nice, full stack, open source solution.

TBA also has an authenticated push API built, which allows a field system to post results in realtime, just sayin' (nudge, nudge ;) )

[SoftwareBug2.0]

My fear is that no news is bad news. If they knew how to fix the problem and how long it would take, they'd probably let us know on TeamForge.

Maybe they can just push the results to Spyder or TBA directly, we both know how to handle the load ;)

Yeah, I mean if things were even "sort of ok" rather than "equipment caught fire" they could release a CSV or something like the output of "mysqldump".

[MrRoboSteve]

I know it's frustrating, but I also know firsthand they've been very busy fixing issues that were blocking match running. We're just going to need to be patient-be assured that they know there's an issue and will work on it at the appropriate priority.

[ronlottem]

Rankings read off by the announcer at the end of the day:

1) 987 101.5 avg
2) 148 81.4 avg
3) 118 77.2 avg
4) 1477 76 avg
5) 4300 (only gave the top 4 averages)
6) 4499
7) 2950
8) 647
9) 704
10) 457
11) 3350
12) 5431

thats what i was looking for actually.
thanks

[plnyyanks]

Updates from FIRST this morning:

Hi Dave,

I understand your frustration. We are working to resolve the issues with the API, but also must focus on supporting the
events at the same time.

The issue we are working through involves the data sync between the local computers at the events and our Azure database
which is the data source for the event web pages and the API.

Regards,
-Matt

...So I would guess we're not getting data this weekend

[Eugene Fang]

Hi Dave,

I understand your frustration. We are working to resolve the issues with the API, but also must focus on supporting the
events at the same time.

The issue we are working through involves the data sync between the local computers at the events and our Azure database
which is the data source for the event web pages and the API.

Regards,
-Matt

I don't get how a data sync issue between events and Azure causes the timeouts/500 errors we're seeing. If there is an issue with data sync, wouldn't the data served by the API just be stale? :confused:

[Ether]

Hi Dave,

The issue we are working through involves the data sync between the local computers at the events and our Azure database which is the data source for the event web pages and the API.

Regards,
-Matt

I hope the data is being safely archived in the local computers, and not lost forever (like the Twitter data in previous years).

[Mike Soukup]

I don't get how a data sync issue between events and Azure causes the timeouts/500 errors we're seeing. If there is an issue with data sync, wouldn't the data served by the API just be stale? :confused:

It would explain why some events have updated rankings, but not matches. It wouldn't explain the massive timeouts & 500 errors. They should be serving empty data. Unless they shut it down today because there were sync problems that only showed up when the DB was under a massive reading load. All speculation.

Until then I'll just reply to all the 1 star ratings for Spyder with a boilerplate "FIRST screwed this up, not us" message.

[Eugene Fang]

Until then I'll just reply to all the 1 star ratings for Spyder with a boilerplate "FIRST screwed this up, not us" message.

Yeah, the ratings are rough :(

[plnyyanks]

No data from Week 1...

http://www.usfirst.org/roboticsprograms/frc/blog-Event-Results-and-API-Data


Event Results and API Data
Blog Date:
Saturday, February 28, 2015 - 13:29

Today’s brief blog is from the FRC software team:

We are aware of tournament data not publishing consistently for all events. The root cause has not been determined, and at this point, we don’t expect resolution for week 1 events. Because of this, developers relying on this data for their own applications will not be able to relay the data on our behalf. This issue only inhibits the data being reported out by events, and the results and rankings reported at events are unaffected. Please accept our apologies for any frustrations.

[Ether]

"developers relying on this data for their own applications will not be able to relay the data on our behalf"

Will the Week1 data remain unavailable even after the datalink problem has eventually been fixed?

"This issue only inhibits the data being reported out by events, and the results and rankings reported at events are unaffected"

But has the data been stored locally, and will it be made available retroactively? Or has the data been permanently lost (like Twitter data in 2014)?

[Swan217]

RoboVisionOD.com has match data for South Florida. I believe we're the only regional to do so.

[magnets]

This is very pathetic that Team 254 mentors can develop a superior field management system in several months outside of their regular jobs while FIRST, which is being paid, fails to have a working system for two years in a row.

The robots are held to a much higher standard than the field software.

[Bryan Herbst]

But has the data been stored locally, and will it be made available retroactively? Or has the data been permanently lost (like Twitter data in 2014)?

The data is stored locally on the FMS.

However, I don't know if there are plans to make that data available to the API.

[Ether]

The data is stored locally on the FMS.

May I ask what is your source for that info?

And is that something new for 2015, or was it also being done in previous years?

[plnyyanks]

May I ask what is your source for that info?

And is that something new for 2015, or was it also being done in previous years?


Ether,

I've been a Scorekeeper in previous years (so this feature may or may not exist this year, but I'd be pretty surprised if it weren't) and FMS would store results locally after every match. It would back up locally and to a flash drive, if one was plugged in. A full backup database would be triggered every 5 matches, or manually by the scorekeeper.

Additionally, at the end of the event, the Scorekepper would export a good amount of data (including match results) to excel format and give it to the FTA. I don't remember if that was standard procedure, or just good practice from the FTAs I've worked with.

I think it's a pretty safe bet the data exists somewhere, but it's a much bigger question if somebody will go through and take all that data off the field manually.

[plnyyanks]

Also, I've created a Google Spreadsheet to crowdsource match results from this week, in case we don't get week 1 data officially.

If you have any data, please enter it here:
https://docs.google.com/spreadsheets/d/1I9tp0QFuWeJM9xBa9FVOqBjcPa2GyVY7cF4UL88O6t4/edit#gid=0

[ATannahill]

Ether,

I've been a Scorekeeper in previous years (so this feature may or may not exist this year, but I'd be pretty surprised if it weren't) and FMS would store results locally after every match. It would back up locally and to a flash drive, if one was plugged in. A full backup database would be triggered every 5 matches, or manually by the scorekeeper.

Additionally, at the end of the event, the Scorekepper would export a good amount of data (including match results) to excel format and give it to the FTA. I don't remember if that was standard procedure, or just good practice from the FTAs I've worked with.

I think it's a pretty safe bet the data exists somewhere, but it's a much bigger question if somebody will go through and take all that data off the field manually.
This is still standard practice. Also standard practice is FIRST using a file transfer system, completely outside FMS, to pull the database from the field to HQ after the event.

[Ether]

standard practice is FIRST using a file transfer system, completely outside FMS, to pull the database from the field to HQ after the event.

Then what does HQ do with that data?

[Bryan Herbst]

Then what does HQ do with that data?




That I don't know. I believe they use that data for stastical purposes (highest scores, average scores, cycle times, etc), and for reviewing match logs if there are any issues that they need to investigate (such as the issues with the field network being compromised at champs a few years back) or problems that are popping up at multiple regionals (say for example if there is a bug in the FMS software causing matches to misbehave).

Again, this is just my speculation as to what they do with the data they pull after events.

[Rachel Lim]

http://frc-events.usfirst.org/ has data for qualification matches, elimination matches, and rankings for some regionals and district events that TBA doesn't. From what I can tell the three come together (if an event has one, it has all three) but I didn't test every single one. Elims have results but the teams in alliances past QFs aren't included.

Those that have data:

Southern Florida: http://frc-events.usfirst.org/2015/FLFO/qualifications
Lake Superior: http://frc-events.usfirst.org/2015/MNDU/qualifications
Northern Lights: http://frc-events.usfirst.org/2015/MNDU2/qualifications
Palmetto: http://frc-events.usfirst.org/2015/SCMB/qualifications
Dallas: http://frc-events.usfirst.org/2015/TXDA/qualifications
FiM Howell: http://frc-events.usfirst.org/2015/MIHOW/qualifications
FiM Southfield: http://frc-events.usfirst.org/2015/MISOU/qualifications
FiM Standish: http://frc-events.usfirst.org/2015/MISTA/qualifications
PNW Aubern: http://frc-events.usfirst.org/2015/WAAMV/qualifications
IN Indianapolis: http://frc-events.usfirst.org/2015/ININD/qualifications
MAR Hatboro-Horsham: http://frc-events.usfirst.org/2015/PAHAT/qualifications

Those I can't find data for:

Inland Empire
Georgia Southern
NE Granite State (schedule but no data)
NE Waterbury (schedule but no data)
PNW Oregon City


This has been the best source for match data (and the only one that comes directly from FIRST) that I've found so far.

[plnyyanks]

TBA implements the broken API and turned off its queries this weekend to try and ease the load on FIRST's server and to avoid wasting CPU time/bandwidth when the API would clearly not give back any data. We'll work on getting the data that is available up soon, either manually or by updating the html parsers for this year's format.

Edit: or by reactivating the API, which appears to be working again...

[Eugene Fang]

All the data that is available through frc-events.usfirst.org is in TBA now.
TBA also has playoff alliances past the quarterfinals.

http://www.thebluealliance.com/event/2015flfo
http://www.thebluealliance.com/event/2015mndu
http://www.thebluealliance.com/event/2015mndu2
http://www.thebluealliance.com/event/2015scmb
http://www.thebluealliance.com/event/2015txda
http://www.thebluealliance.com/event/2015mihow
http://www.thebluealliance.com/event/2015misou
http://www.thebluealliance.com/event/2015mista
http://www.thebluealliance.com/event/2015waamv
http://www.thebluealliance.com/event/2015inind
http://www.thebluealliance.com/event/2015pahat

[Ether]

This is still standard practice. Also standard practice is FIRST using a file transfer system, completely outside FMS, to pull the database from the field to HQ after the event.

Does this mean that somewhere at HQ is a copy of the complete Twitter data from 2014, including all the data that was missing because of datalink problems at events?

[Mark McLeod]

Does this mean that somewhere at HQ is a copy of the complete Twitter data from 2014, including all the data that was missing because of datalink problems at events?
No recorded Twitter feed. It would be final match results as you'd normally find on the FIRST website.

[Eugene Fang]

No recorded Twitter feed. It would be final match results as you'd normally find on the FIRST website.

So no auton/teleop/assist/truss/etc. breakdowns?

[Ether]

So no auton/teleop/assist/truss/etc. breakdowns?

That's exactly why I was asking: auton/teleop/assist/truss alliance scores per match, not a total sum of alliance score categories per team.

[MikeE]

Does this mean that somewhere at HQ is a copy of the complete Twitter data from 2014, including all the data that was missing because of datalink problems at events?


Based on my limited understanding of the system I believe the twitter data was generated on-the-fly, so it's unlikely that the it exists separately anywhere.
However you're really looking for a breakdown of the match score. Twitter was one of the two public sources providing score component level data, the other being audience result screens (which are somewhat less convenient to parse!).

I do know that component scores are part of the event database but not if this level of detail is maintained after the season.
You would have to ask FRC HQ if that data is still available, but obviously that's going to be a low priority compared to their critical work on this season.

[Ether]

Based on my limited understanding of the system I believe the twitter data was generated on-the-fly, so it's unlikely that the it exists separately anywhere.

Maybe I'm just an optimist, but it seems to me that the designers of the FMS scoring database would have saved all the alliance scores for each match (i.e. the Twitter data), not just the final total scores. We're not talking gigabytes here.

However you're really looking for a breakdown of the match score.

Correct.

Twitter was one of the two public sources providing score component level data, the other being audience result screens (which are somewhat less convenient to parse!).

Could you elaborate what you mean by audience result screens? I think you are joking, but just want to make sure.

I do know that component scores are part of the event database but not if this level of detail is maintained after the season.

That's basically what I am asking, and getting conflicting responses.

You would have to ask FRC HQ if that data is still available,

Been there, done that, never got a response.

[Mark McLeod]

So no auton/teleop/assist/truss/etc. breakdowns?
Sorry, it does include those breakdowns.
The database, a priori, includes everything necessary for the Head Ref/Scorekeeper to be able to review and correct the different scoring elements of previous matches.

When a team stands in the Question box and politely discusses the previous match scores with the Head Ref, the database has all the information required.

[Ether]

Now I'm confused. What's the difference between

So no auton/teleop/assist/truss/etc. breakdowns?

and

Sorry, it does include those breakdowns.
The database includes everything necessary for the Scorekeeper/Head Ref to be able to review and correct the different scoring elements of previous matches.

[Mark McLeod]

I don't understand the question either :-)
So my phrasing is deficient.

[Joe Ross]

I think the difference is between the data being potentially available, and FIRST spending the money to make it available, especially considering all the other things that they are working on.

[MikeE]

Could you elaborate what you mean by audience result screens? I think you are joking, but just want to make sure.


Only half joking.

The only public sources for component level scores in 2014 were the twitter feed and the results screen displayed to the audience at the end of every match.
Where high enough quality video exists of a match, especially recordings of the webstream, then the component level scores are available as images within the video, at least in principle.
It is clearly a much more difficult process to extract the scores from an image than from text (this is the "joking" part). However a sufficiently motivated individual, or perhaps community crowd sourcing, could extract the data.