• Resolved dpitzer

    (@dpitzer)


    Hello,

    We are working on a larger site and box and we have run into an issue converting the wp_usermeta to the optimized form. All the other tables have converted correctly. This table has 1,097,079 rows, approximately, but this is way smaller than the wp_postmeta (4,349,147 rows, approximately) which worked.

    We are working in a staging environment, so if you would like to see anything, please just ask.

    Index WP MySQL For Speed 1.4.4
    Versions Plugin:1.4.4 MySQL:8.0.27-18 WordPress:5.9.3 WordPress database:51917 php:7.4.29
    Fatal error: Uncaught ImfsException: [0]: MySQL server has gone away in ALTER TABLE wp_usermeta ADD UNIQUE KEY umeta_id (umeta_id), DROP PRIMARY KEY, ADD PRIMARY KEY (user_id, meta_key, umeta_id), DROP KEY meta_key, ADD KEY meta_key (meta_key, meta_value(32), user_id, umeta_id), ADD KEY meta_value (meta_value(32), umeta_id), DROP KEY user_id/**imfs-query-tag*986555825*/

    thrown in /var/www/staging.undergroundreptiles.com/htdocs/wp-content/plugins/index-wp-mysql-for-speed/code/imsfdb.php on line 375
    Error: There has been a critical error on this website.Learn more about troubleshooting WordPress. There has been a critical error on this website.

    Thank you,

    Dave

Viewing 4 replies - 1 through 4 (of 4 total)
  • Plugin Author OllieJones

    (@olliejones)

    Hi, sorry you’re having difficulty. This is the heart of your error message:

    MySQL server has gone away

    This means your MySQL server crashed and restarted in the middle of rekeying the wp_usermeta table. Do you have many users? With lots of metadata each?

    The rekeying operation requires temp space on your hard drive, enough to hold the table being rekeyed with its indexes. MySQL really doesn’t like it when the temp space runs out, and restarts. And, on some Linux installations the /tmp directory is a ramdisk to make things fast, so it runs out quickly.

    So, try tweaking your MySQL configuration. See this.

    https://dev.mysql.com/doc/refman/8.0/en/server-options.html#option_mysqld_tmpdir

    As root on your MySQL machine edit the /etc/mysql/mysql.conf.d/mysqld.cnf file.

    Near the line where it says datadir = something/or/other add a line saying

    tmpdir = /var/tmp

    then restart your server by saying

    sudo service mysql restart

    Please let me know if this was the problem.

    By the way if you do have many users check out the companion plugin. https://www.ads-software.com/plugins/index-wp-users-for-speed/

    Thread Starter dpitzer

    (@dpitzer)

    Thank you very much for your feedback on this.

    The tmp space free is 300GB and the DB is 3 GB. We did find that we needed more space in our innodb buffer pool and log file size, but even after that it still fails.

    We did add an explicit setting for the tmp dir.

    The companion plugin / index has been installed and worked fine. Can you send me the SQL it is running and I can run it inside tables plus?

    Thank you,

    Dave

    Plugin Author OllieJones

    (@olliejones)

    Thanks for trying the /var/tmp fix; sorry to hear it didn’t help your problem.

    Would you be willing to upload your site’s metadata so I can see it? If so, please visit

    Dashboard > Tools > Index MySQL > About

    and click the Upload Metadata button. I’ll take a look. Also, take a look at the last hundred or so lines of MySQL’s log file. Maybe it explains the crash. On Ubuntu it’s at /var/log/mysql/error.log .

    The companion plugin isn’t like this one. This one rekeys tables, and so has some straightforward SQL statements you can run just once. Your original question showed the entire once-and-done SQL statement that rekeys the wp_usermeta table.

    That one (https://www.ads-software.com/plugins/index-wp-users-for-speed/) adds rows to wp_usermeta to get away from an inefficiency in WordPress Core’s way of looking up administrator, editor, and other users by role. That inefficiency doesn’t matter in sites with small numbers of users, but it does when you have more users. It’s necessary to add those rows in batches to keep the transaction sizes reasonable (a SQL statement that INSERTs tens of thousands of rows at once can blow out your logs and/or make your system thrash). Then, that plugin hooks WordPress’s user queries and replaces them with ones that use the added rows. Upshot: that plugin doesn’t have clean SQL you can run elsewhere.

    Plugin Author OllieJones

    (@olliejones)

    Try running the query that threw the “MySQL server has gone away” error from a MySQL client like phpmyadmin or the command-line mysql. That should update your wp_usermeta indexes. This is the query.

    ALTER TABLE wp_usermeta
      ADD UNIQUE KEY umeta_id (umeta_id),
      DROP PRIMARY KEY,
      ADD PRIMARY KEY (user_id, meta_key, umeta_id),
      DROP KEY meta_key,
      ADD KEY meta_key (meta_key, meta_value(32), user_id, umeta_id),
      ADD KEY meta_value (meta_value(32), umeta_id),
      DROP KEY user_id;

    All the queries are here.

    If it gets the same “MySQL server has gone away” error you should look at the last hundred lines or so of your MySQL error log. It may tell you what went wrong. This query tells you the location of that log on your MariaDB / MySQL server machine.

    SHOW VARIABLES LIKE 'log_error';

    I hope you can figure this out; your MySQL server seems to be unstable, and that’s not good.

Viewing 4 replies - 1 through 4 (of 4 total)
  • The topic ‘Fatal Error Uncaught Exception’ is closed to new replies.