Noticed today a lot of the Character and Anime view counters were slowing down the site.
Especially when new characters are added, it can take an average of 1 second to rebuild the indexes. A lot of the errors tho were for cases taking around 10 seconds to complete.
While I would like to code an extendable database framework in C, it's a project that should take upwards of a week, and one I'd like to do concurrently with a Binary Tree module ... which I still lack.
Anyways, the Views are expendable. They generate themselves, after all. THat's why I can code a risky replacement which could lose all data. And it's why I can change the storage engine from InnoDB to memory. I had tried different indexes, but it was kinda like an idiot pressing all the buttons on a remote to change the channel. I did more harm than good. The memory storage engine should never take more than a second to complete any operation of this fixed data set. Even without proper indexes.
And another update on the stats server memory usage.
Seems to have stabilized now.
73375 rei 1 45 1 19112K 15608K accept 1 113:36 0.00% ss
19MB max allocated, and 15MB in use. Not bad at all.
Hopefully later this month I can start on the next version of the Stats Server.
Feature requests include:
- Binary Tree Indexes
- export to disk
- load from file
- dynamic table frameworks
Started work on the Guide Books project.
It's something fun and easy, but led off on a difficult tangent.
This is roughly the 500th time I've created an HTML table. And likely the 300th time I'll write handlers for FORM data. Seems like as good a time as any to analyze all of them, find the similarities, and write a dynamic generation application to cut down on the duplication.
So working on that...
Another memory usage update on the session server.
73375 rei 1 45 1 19112K 15596K accept 3 72:23 0.00% ss
It's nearing 20MB of allocated heap. Wonder when/where it'll hold. 27M to 30M might not be so bad.
I've also been optimizing various MySQL queries, correcting table definitions from back when I was still a Noob. Removing useless indexes, and adding crucial JOIN indexes. Plus completely rewriting some queries so they examine less data and execute more quickly.
mysql> explain SELECT COUNT(*) AS n FROM (SELECT author,COUNT(*) AS n FROM characters GROUP BY author) AS a WHERE n > 10;
+----+-------------+------------+------+---------------+------+---------+------+-------+---------------------------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+------------+------+---------------+------+---------+------+-------+---------------------------------+
| 1 | PRIMARY | | ALL | NULL | NULL | NULL | NULL | 326 | Using where |
| 2 | DERIVED | characters | ALL | NULL | NULL | NULL | NULL | 25895 | Using temporary; Using filesort |
+----+-------------+------------+------+---------------+------+---------+------+-------+---------------------------------+
2 rows in set (0.02 sec)
mysql> create index author on characters(author);
Query OK, 25895 rows affected (0.80 sec)
Records: 25895 Duplicates: 0 Warnings: 0
mysql> explain SELECT COUNT(*) AS n FROM (SELECT author,COUNT(*) AS n FROM characters GROUP BY author) AS a WHERE n > 10;
+----+-------------+------------+-------+---------------+--------+---------+------+-------+-------------+
| id | select_type | table | type | possible_keys | key | key_len | ref | rows | Extra |
+----+-------------+------------+-------+---------------+--------+---------+------+-------+-------------+
| 1 | PRIMARY | | ALL | NULL | NULL | NULL | NULL | 326 | Using where |
| 2 | DERIVED | characters | index | NULL | author | 4 | NULL | 25895 | Using index |
+----+-------------+------------+-------+---------------+--------+---------+------+-------+-------------+
2 rows in set (0.01 sec)
TOP 
News
Log In
