The New Sudoku Players' Forum

[dobrichev]

Code: Select all

1..2....3
.4..5..1.
..6...7..
8....1.3.
.3..4.6..
...9.....
..5.7....
.1.3...72
9......8. ED=10.0/10.0/10.0


[Pat]

after announcing the game-start
and before adding the matching post on Results --

• expected: 10-minute delay (how?)
• observed: crash (+ automatic restart after 5-minute delay)

sorry i'm hung up on this silly question,
but HOW does one launch a 10-minute delay ?


the crash happens



• before ?
  
• during ?
  
• after ?
this launch

[JasonLion]

g.r.emlin is mostly a collection of shell scripts. When there is something to post it checks the last post time, and if it was too recent it sleeps for a while using the Linux sleep command. Previous post time is lost on a restart.

So far I have found that the ratings database is corrupt. I have not found anything that could have corrupted it, nor anything that will rebuild it. I also don't appear to have a backup of a valid database (several backups of the bad one). I am fairly sure I could reset the database to empty, but then it would forget what ratings it has seen in the past. Glenn should have a backup of a valid database, he may also know how to rebuild it, but he has not been able to find enough time to look at things.

The crash is very early in the process of building the post for the results topic. It goes to load the ratings database, to report on ratings seen in the current game, and crashes. The sleep, if it got that far, would be after the post was completely written and before it was posted to the forum.

[dobrichev]

Hi Jason,

Thank you for the efforts.

Maybe it is time to think about rewriting from scratch?

The aggregated data should be reproducible from the raw data, i.e. from the posts.
Is the corrupted DB a set of textual files? If so, then you could try the LC_ALL=C environment variable, or other locale setting(s) that is sorting the text and dates fields in the expected order.

Do the current scripts communicate directly with the forum software (DB or API), or they scan the HTML posts in a way that any user's browser scans them?

[JasonLion]

g.r.emlin uses HTTP to talk to the forum software, just as a web browser would, parsing select bits of the HTML (login cookies and input forms mostly) and filling out the new post form much as a real person would.

It is clear that all of the information is there to rebuild the ratings database, but I don't see any code to do so without re-scoring (and posting about) all of the old games. Plus, there is still the mystery of what went wrong with the ratings database in the first place.

[Pat]

So far I have found that the ratings database is corrupt.

I have not found anything that could have corrupted it,
nor anything that will rebuild it.

I also don't appear to have a backup of a valid database
(several backups of the bad one).

The ratings history file got hosed in the transfer.

That's restored and the referee is running again.

[Pat]

the leader (and most-likely-winner) is dobrichev

meaning: very-likely

any other player would need a 10.7 (or higher)
to make the race interesting

[JasonLion]

::: status :::
With a little luck, we may be working again

[Pat]

we need the check-point back at 254852

[JasonLion]

It is trying to display the status. Never seems to finish however. I'm going to give it a long time.

[g.r.emlin]

::: adjustment 2017-02-04+23:34:00-0000 JPF not-minimal -50 2017-02-05+13:31:00-0000 JPF not-minimal -50 :::

[g.r.emlin]

::: game 0280 status LIGHTNING 2017-02-13+16:34:47-0000 :::

Code: Select all

game 0280 2017-02-03+16:36:00-0000 LIGHTNING duration 1w02d dealer Pat pages 2204-2209 (open) 4 players 2 resigned
_____________________________________________________________________________________________________________

    2.0/1.2/1.2 Pat            4.8/1.2/1.2 dobrichev      6.5/5.0/2.6 dobrichev   10.0/10.0/10.0 dobrichev
    4.2/4.2/4.2 dobrichev      5.8/3.0/2.8 Pat            6.9/6.9/3.0 dobrichev    10.1/10.1/9.4 dobrichev
    4.4/4.4/4.4 dobrichev      5.9/2.8/2.8 JPF            7.5/7.5/7.5 dobrichev    10.3/10.3/9.6 dobrichev
    4.5/4.5/3.4 JPF            6.0/1.2/1.2 dobrichev      7.9/7.9/7.2 JPF
    4.6/4.6/3.4 dobrichev      6.1/1.2/1.2 Pat            8.0/8.0/6.7 JPF
    4.7/4.7/3.4 m_b_metcalf    6.2/6.2/3.4 JPF            9.6/9.6/9.5 JPF
_____________________________________________________________________________________________________________

    open ratings : 1.2 1.5 1.7 2.3 2.5 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.9-5.2 5.4 5.6 5.7 6.6-6.8 7.0-7.4 7.6-7.8 8.1-9.5 9.7-9.9 10.2 10.4-11.4 11.6-11.8

       submitter  entries  score  response  notes
     -----------  -------  -----  --------  -----
     m_b_metcalf        1     34     3h02m  pearl,resigned
             Pat        3    160    22h59m  rare-1,last-entry=63,consecutive-entry,dealer
             JPF        6    319     9h29m  pearl*5,not-minimal*2,resigned
       dobrichev       11    898     1h01m  rare-1,diamond*4,pearl*4,highest-rating=80
     -----------  -------  -----            -----
               4       21   1411            rare-1*2,diamond*4,pearl*10,resigned*2


[g.r.emlin]

::: rarity :::

Code: Select all

18 100200003030040050006000400200005060050020700000800000003070000070500024800000070 13m00s rare ED seen 5 times before
37 100200003040030050006000700200001060080050200000800000005010000070300019800000040 13m00s rare ED seen 6 times before
23 100200003040050010006000700500006020080020600000700000003090000010400058200000030 16m00s rare ED seen 7 times before
35 100200003040050060007000100800009010020060400000700000005070000060500032900000080 26m00s rare ED seen 8 times before
74 500100003070060040008000500900008030060010400000700000009020000010400069300000080 1h23m rare ED seen 9 times before
2017-02-05+18:19:00-0000 6.5/5.0/2.6 dobrichev rare ED seen once before
2017-02-13+13:55:00-0000 5.8/3.0/2.8 Pat rare ED seen once before


[Pat]

trouble !!!

it posted the status OVER the game-start
at 254852

[Pat]

status now shows

9.6/9.6/9.5 JPF

as puzzle # 2

it is in fact # 60 or 61

i.e.
missing a huge chunk at start