Gah, I managed to trigger a weird bug in Kareha that will wipe threads if the disk is full and you try to rebuild caches. And my last backup of the board is totally ancient. I don't know if there are any other backups elsewhere. Does anyone else happen to have any more recent copies of threads?
PS: Don't bother posting in any other threads than this, the posts will most likely get wiped if I can find some newer data.
Something else had gone out of control and used up the space.
Well, I'm glad to know I'm not the only one whose codebase doesn't handle that properly.
Well, I made a script to reconstruct threads and metadata from the Google cache. This is about as much as I can do to fix it. It looks like it's gone fairly well - a week or two of replies seem to be missing, but there hasn't been much action lately anyway.
If you posted something important that has been wiped, feel free to repost!
In case anyone else ever needs it, I made put up the script I used here: http://wakaba.c3.cx/releases/other/recover.pl
To use it, first use wget to grab archives out of the Google cache. I used a command lines like:
wget -r -l1 -U Mozilla/4.0 -e robots=off -H -D 72.14.207.104 'http://www.google.com/search?q=site%3Awakaba.c3.cx%2Fsoc%2F&num=100&start=0'
wget -r -l1 -U Mozilla/4.0 -e robots=off -H -D 72.14.207.104 'http://www.google.com/search?q=site%3Awakaba.c3.cx%2Fsoc%2F&num=100&start=100'
wget -r -l1 -U Mozilla/4.0 -e robots=off -H -D 72.14.207.104 'http://www.google.com/search?q=site%3Awakaba.c3.cx%2Fsoc%2F&num=100&start=200'And so on. Next, if you have an old backup, put it in the current directory. Then, run the script like so:
perl recover.pl path_to_google_archives/*And it should do its best to rebuild those threads that are newer in the Google cache files than in your own backup.
I cut-and-pasted some posts that were in the cache of the frontpage, which my script doesn't handle.
I also noticed that I lost a lot of Post deleted by moderator from spammers! Hooray!
What happened to http://wakaba.c3.cx/soc/kareha.pl/1100499906/
(Tripcode Decoder)?
Good question. I restored it again, and fixed some weirdness in the header that shouldn't have caused problems but maybe it did anyway.
Let's see if it sticks around this time.