Litany of Database Recovery

by Kieran Healy on June 20, 2004

Queen of SQL statements. _Pray for us_.
Empress of Emacs. _Pray for us_.
Sacred Heart of Search and Replace. _Pray for us_.
Defender of Write Permissions. _Pray for us_.
Patron of Manually Edited Dump Files. _Pray for us_.
Savior of unexpectedly small Disk Quotas. _Pray for us_.
Shepherd of Lost Posts. _Pray for us_.
Protector of Hapless Administrators. _Pray for us_.
Scourge of “Wholly Inadequate Import/Export Formats”:http://lemonodor.com/archives/000730.html. _Pray for us_.
Mother of all the Bloggers. _Pray for us_.

I think we’re back. Fresh — or at least unrotted — permalinks and all. Thank you, thank you to everyone who commented in the now-destroyed post where I wailed about the problem. The solution was to get an SQL dump of the database from the old server and read it in to the new database. Not as easy as it might have been, because the old server had old blogs, with old hard-linked archive sources and all the rest of it. But I think it worked.

My sincere apologies to fellow-posters and commenters whose recent contributions got deleted in the course of the database restoration. I guess I revealed myself to be a utilitarian at root: five or six posts and their comments were sacrificed on behalf of about two thousand posts and their permalinks. Moral: Do not put me in charge of interrogating suspected terrorists.[1]

So, as predicted in my “Don’t Upgrade”:https://www.crookedtimber.org/archives/001967.html post, I’ve just spent an unconscionable amount of time (I am about to start paying off large debts to my wife and daughter) getting us back to where we were last week. But now we are where we were last week, but on new servers. To switch religions momentarily, Oy.

fn1. Alternative moral for high-ranking Pentagon officials: By all means put me in charge of interrogating suspected terrorists, because I will lose them.

{ 1 trackback }

b6bd0fd7c9dbda9b
04.21.05 at 12:37 pm

{ 20 comments }

1

belle 06.20.04 at 2:19 pm

Thank you very much, Kieran. Hey, you know how I posted right in the middle of the move, like a big retard, and now my post is the only one left? Isn’t that ironic?

2

q 06.20.04 at 2:32 pm

First rule of computer software administration: don’t let other people take you for granted: make the odd critical unforeseen error to drive home the issue. It is funny how many systems start going wrong shortly before the IT staff are due to leave!

3

Chris Bertram 06.20.04 at 2:45 pm

Fantastic job Kieran, well done!

4

Dave 06.20.04 at 3:25 pm

Congrats, dude. I can finally start linking you again.

5

yabonn 06.20.04 at 5:52 pm

Mother of all the Bloggers. Pray for us.

My blogline doesn’t upagrade anymore, by i’m sure it’s normal, and that it’s no real problem, as long as the thirty six goats were properly sacrificated.

Eat, Moloch! Eat!

6

maria 06.20.04 at 6:11 pm

Thanks a million Kieran, you’r a total superstar and I promise if I ever meet you, I’ll buy you a pint!

7

David Tiley 06.20.04 at 6:13 pm

This is the Weird of the internet. You neglect your wife and daughter – flesh and blood and the smell of love in the morning – for the sake of thousands of people around the world that you have never met.

Well..

Thank you.

8

Ophelia Benson 06.20.04 at 7:51 pm

“You neglect your wife and daughter – flesh and blood and the smell of love in the morning – for the sake of thousands of people around the world that you have never met.”

Not all that weird. Not weird at all really. People do that any time they work on a book, for example, when they might possibly be interacting with a living human being instead. Interacting with living human beings is good but it’s not the only good.

(Of course, fiddling with irritating techy stuff is a different sort of thing from writing a book…but then there’s irritating stuff connected with writing a book, too. Notes, index, bibliography, all that sort of thing. So the point remains.)

9

Ophelia Benson 06.20.04 at 7:57 pm

By the way, Kieran – a few of the lost posts are still visible via Google’s cache, you know – so I assume you could copy and re-post them as if they were new ones?

10

todd. 06.20.04 at 7:58 pm

Any man who prays to the Empress of Emacs and the Sacred Heart of Search and Replace probably needs to spend a few hours banging away on such a problem every now and again, just as badly as he needs the smell of love in the morning.

Well, maybe not just, but badly nonetheless.

11

yabonn 06.20.04 at 8:33 pm

Tried to re syndicate, but bloglines still believes that there are no new articles on ct.

.. Thirty six goats, and the NewsFeed Sacred Dance should do the trick.

12

John Quiggin 06.20.04 at 10:23 pm

Order is restored! But where has all the entropy gone?

13

Ophelia Benson 06.20.04 at 10:58 pm

Well there, that’s what I said, those are the ones Google cached. Except for the Microsoft one, that wasn’t there, so it must have been some other magic.

As for the entropy, that’s still there, somewhere between the incompleteness and the uncertainty. Not too far from Copenhagen.

14

yabonn 06.21.04 at 12:12 pm

Sooooo, erm, i gather i’m alone having that problem that my bloglines aggregator doesn’t get new posts from ct, right?

15

Kieran Healy 06.21.04 at 1:13 pm

Sooooo, erm, i gather i’m alone having that problem that my bloglines aggregator doesn’t get new posts from ct, right?

No. My CaRP aggregator on my homepage isn’t updating either, and I don’t know why — NetNewsWire has no trouble.

16

yabonn 06.21.04 at 2:23 pm

No. My CaRP aggregator on my homepage isn’t updating either,

Relief.

Theory : bloglines registers the url.xml, then uses the ip adress for post retrieval. If so they’ll have to update on their side.

I don’t know carp, maybe a similar glitch explains that case too.

17

Kieran Healy 06.21.04 at 10:34 pm

Theory : bloglines registers the url.xml, then uses the ip adress for post retrieval. If so they’ll have to update on their side.

I don’t know carp, maybe a similar glitch explains that case too.

This is the only explanation I can think of but it doesn’t make any sense. Carp is a little php program and it’s not caching ip addresses anywhere I can see. And it would be just insane if bloglines used the IP address of sites — avoiding that is the whole point of DNS servers.

18

yabonn 06.22.04 at 12:37 am

… but it doesn’t make any sense.

Yup. Beats me.

Shooting in the dark, and as we are deep in weirdness : would it be possible the old ip is stuck somewhere in the mt config file ? Some kind of “go update at this address” setting messing somehow with bloglines/carp?

… Got my geekish feathers all ruffled now. If this is just a dns propagation quirk i’m going to make all kinds of unpleasant noises.

19

Kieran Healy 06.22.04 at 12:59 am

Shooting in the dark, and as we are deep in weirdness : would it be possible the old ip is stuck somewhere in the mt config file ? Some kind of “go update at this address” setting messing somehow with bloglines/carp?

I’m pretty sure that it’s not anything like this. Compare the CaRP feed on my Arizona homepage (not updating) with the otherwise identical CaRP feed on my Blog Host homepage (working fine). One has updated and the other not. My kieranhealy.org is hosted by the same company that now hosts crookedtimber.org. It must be DNS, maybe with some PHP-related cache or other taking an appallingly long time to refresh.

20

yabonn 06.22.04 at 1:41 am

But, but but…

Both are up to “Crooked Timber’s Greatest Hits” on my browser, last comment finishing by “… most popular draws.”

Outdated Ip info stored by the browser then? Tried delete cache, delete bloglines cookie, deep reload, no avail.

Grmbl. Bed.

Comments on this entry are closed.