[wp-hackers] Problems ?

[Jason Hoffman]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mar 6, 2005, at 3:52 PM, Carthik Sharma wrote:

> Thanks for the input Jason,
>
> Let me try and summarize the gist of what you say :
>
> There seem to be problems when
> 1) php's running in cgi mode
> 2) people use external editors + xmlrpc posting

Yes, that's been the people that came here after getting the boot. Then 
the same 25% of a CPU + mysql being crashy.

For example they post and

041123 14:30:31 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041123 14:30:31 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041124 14:27:30 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041124 16:18:59 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041124 16:18:59 Out of memory;  Check if mysqld or some other process 
uses all available memory.

And that's with a fat mysql. So that's on guy that was booted from 
lunarpages.

Then same guy  and a comment spam victim

041217 15:23:20 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041217 15:23:20 read_next: Got error 12 when reading table 
./powlon108/wp_posts
041217 17:43:13 read_next: Got error 12 when reading table 
./bsag-wp/wp_comments
041217 17:43:13 read_next: Got error 12 when reading table 
./bsag-wp/wp_comments
041220 15:43:51 /usr/local/libexec/mysqld: Out of memory. Restart 
daemon and try again (needed 2097144 bytes)
041220 15:43:51 /usr/local/libexec/mysqld: Sort aborted

And then someone familiar whose table we repaired (don't know if he 
uses an external editor)

050123 16:51:55 read_next: Got error 12 when reading table 
./powlon108/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050124 19:36:43 read_next: Got error 12 when reading table 
./rboren-weblog/wp_posts
050125 01:42:54  mysqld started

Actually all of our read_next errors and mysqld crashes are on wp_posts 
(#1), wp_comments (#2) and phpbb's  search (that's different, it's  
Can't open file: 'search_matches.MYI'. (errno: 145) ).

This happens in 4.0.22, 4.0.23, 4.1.7-4.1.10, and an error 12 is 
"System error:  12 = Cannot allocate memory". Now this is mysql that's 
tuned up pretty high, and the most common place one normally sees error 
12 is in bdb and innodb when you run out of locks or large 
multiple-statement transactions ...

So I would definitely look at your query as well.



> Maybe 2) needs to be looked into, any details about how to recreate
> the situation that cause tables to go bad after posting with such
> clients would be great.
>
> One question : What kind of repair do you recommend? Just fixing the
> indexes, or a more elaborate repair? Is there something all WordPress
> users could do with, on a periodic basis that could perhaps go into
> the upgrade.php script or some such? (Alright, that is more than just
> one question ;) )

We just run a repair table and have a script that checks for bad table, 
locks them and runs a myisamcheck repair

- - J

-----BEGIN PGP SIGNATURE-----
Version: PGP 8.1

iQA/AwUBQiueHlUyB+ajXkCLEQKdLwCg9yE8LW6XFKc7oLzRQWrTF/BJLaQAnjUB
0grRfKCySZSuYoU2cQ7fOmip
=rUJa
-----END PGP SIGNATURE-----