[reader50]

I'm expecting this to happen Thursday evening PST. If the plan changes, Scott or myself will post an update. With luck, the outage will last an hour or so.

Predictor code is going up, so that stats will work once the project comes back. Sigs will work right away.

Assorted files throughout the site need to be replaced during the update, to keep the rest of teamstats code in sync with the Predictor versions. Otherwise something would break for sure.

If something breaks anyway, we might be down for a bit longer than an hour. Life would get boring if these things didn't happen unexpectedly.

[Scotttheking]

And if something major breaks, reader50 will have to wait until Sunday when I theoretically have some free time.

[reader50]

Teamstats is back up, with Predictor enabled. Predictor data won't appear right away, there will be no data until a stats pull happens. If it hasn't done a stats run by morning, I'll force an update.

The entry page and nav bars no longer mention dFold or Ubero, but both are still present by going to the URLs. If there are no objections, I'll probably nuke the old dFold1 section, it's pretty dead.

If any glitches turn up, please post them here. Nearly every teamstats file got replaced, and something may have slipped through.

[reader50]

Predictor stats are not showing as available. I ran the scripts manually, fixed some bugs, and ran them again. Still no data showing, even though it's there in mysql.

Some poking around revealed that the DB box's clock is set approx 6 hours ahead of the web server box. Until the webserver catches up to the DB box's timestamp, it will ignore data from the future. So stats will appear in a few more hours.

Naturally, I figured that out about 5 minutes after Scott left for the weekend.

------------
For the technically interested, this matters only on Predictor. It is the first project that no longer uses any scratch files on disk for project-summary data. Since those scratch files were also used by the QuickStats bars, it meant the data parsers had to be on the webserver box, even though the processing scripts had long since been moved to the DB box.

We've been trying to move all the backend scripts to the DB box in order to make the webserver respond faster to page requests, and the parser scripts are the last ones left on the webserver. With Predictor using the DB for all data storage, it's parser is on the DB box too, and thus, the data got timestamped using the DB box's clock.

I could fix it by wiping the existing data entry and moving the Predictor parsers back to the webserver. If Predictor were running today, that would be the fast answer. Since the data is static until the project comes back, it's simpler to let the data appear in a few hours, and wait for Scott to get back to reset the DB server's clock in a day or three.

[mikkyo]

Yay!
The Predictor stats are showing up now, and it looks like everything is working with the stale data.
I look forward to the cloning for CP and Seti.
Good job reader50!