I can ping the new server! (in General)

AdminJonathan March 30 2007 2:01 AM EDT

I got some more prep work done while I was stress testing it post-repair, but I still have probably a day's worth of work and testing before I can load the db and call it ready. Monday should be a pretty safe bet though!

Meantime, I'll drop the DNS TTL to 1h so when we do switch over we don't have to wait 8h+ for everyone's ISP to see the new one.

QBRanger March 30 2007 2:09 AM EDT

Good news to hear!

QBRanger March 30 2007 2:11 AM EDT

One Q:

How long will cb be down while we change to the new server?

[T]Vestax March 30 2007 2:26 AM EDT


bartjan March 30 2007 2:38 AM EDT

PM: probably a very short time. But then the few hours between the moment the database was copied over and the moment the DNS switches over will not be saved.
At least, that's what has happened the previous hardware upgrade.

AdminNightStrike March 30 2007 2:56 AM EDT

What will happen to auctions during the outage?

bartjan March 30 2007 3:00 AM EDT

It's probably not a smart idea to have your auctions end at monday...

Popsicle Man Supreme March 30 2007 3:21 AM EDT

what would happen if PM's char magically disappeared in the transition. :-)

[RX3]Cotillion March 30 2007 3:33 AM EDT

We'd all be that much closer to the next level of BA nerfage? *grins*

AdminJonathan March 31 2007 12:32 AM EDT

Wow, postgresql 8.1's dump is finally smart enough to figure out all the circular relationships in CB. For the first time I haven't had to manually massage the dump before it could be loaded into the new db. Bravo!

QBsutekh137 March 31 2007 1:04 AM EDT

"Massage the dump"...

And folks thought, "tanks shouldn't be able to use death to dodge" was one for the books... I think Jonathan has that beat, hands down. Or, hands massaging, as the case may be.

Letting PostgreSQL massage the dump is bravo-worthy indeed! *smile*

AdminJonathan March 31 2007 1:17 AM EDT

I have to admit that made me laugh. :)

QBsutekh137 March 31 2007 1:19 AM EDT

In all seriousness, do you have a quick example of a circular relationship that it used to have trouble with? How would you massage, just by changing the script and altering in a different order after the main parts were loaded? Hooray for data talk!

AdminJonathan March 31 2007 12:25 PM EDT

yeah, it used to get confused with the dependencies... for instance
CREATE TABLE minion_spells (
    minion_id integer NOT NULL,
    spell_id integer NOT NULL,
    "level" integer NOT NULL,
    CONSTRAINT spell_type_c CHECK (spell_ok_p(minion_id, spell_id))
it would put the definition of spell_ok_p after the table creation, which didn't work too well. Possibly in 7.x postgresql you couldn't create a function if a table it referenced didn't exist yet; I don't remember for sure. More likely the dependency calculation was just screwed up.

QBsutekh137 March 31 2007 2:31 PM EDT

Such a tight database design! Here I thought you were just making it all up as you went along... *smile*

Yeah, database engines seem to have trouble with things that are more "compile"-like operations involving declarations and such... PostgreSQL must have added a decent parser to run scripts through before actually running each atomic statement That's slick.

Thanks for elaborating! Now go supervise your dump massage! *smile*
This thread is closed to new posts. However, you are welcome to reference it from a new thread; link this with the html <a href="/bboard/q-and-a-fetch-msg.tcl?msg_id=0024Ij">I can ping the new server!</a>