Skip to main content

My mysql server does not restart

Comments

24 comments

  • R A L

    We have the same issue!! It needs to be fixed asap. Customers are down, and we have more clients on other servers. We don't want this to happen on all our other servers either.

    0
  • Tony

    Paying premium for WHM and they neglect basic update security. Not good for anyone, loosing money and reputation.

    0
  • Jeff

    What an absolute debacle. Luckily it had only been updated on one of my servers. But still, extremely frustrating that there's no checks and balances. And here I am troubleshooting a faulty server update at 1am. Boy, there's going to be a lot of upset website owners and server administrators when they wake up in the morning (US) and discover that their mysql was updated without their consent.

    Sure am glad I pay such a premium for vetted software updates. /s

    0
  • Eduard Dragan

    Any update on this?

    cPanel guys published 2 workarounds here: https://support.cpanel.net/hc/en-us/articles/39925555560471-MySQL8-4-upgraded-to-MySQL9-7-during-nightly-updates

    Unfortunately neither one works for me for some reasons.

    The server is down since last night and we have a lot of disgruntled customers.

     

     

    0
  • Pablo Gómez

    Same here. Server doesn't start, but following the procedure on the linked page for that case doesn't allow me to restart it anyway. Looks like databases were converted to 9.7 before server crashed. The error in logs is "Failed to initialize DD Storage Engine".

    0
  • Alan Gilson

    I'm one of the fortunate ones in that the fix worked for me. There is hope.

    0
  • Dan Muey

    I completely understand how frustrating this has been — when MySQL is unexpectedly down, it’s disruptive and stressful, especially on a production system.

    I do want to share some context that may help: this issue was not specific to cPanel or Plesk. It stemmed from an upstream problem in the MySQL 8.4 packages themselves, and it affected everyone running that version on operating systems that received the bad update, regardless of control panel or platform.

    We’d strongly encourage participating in the upstream MySQL bug report so the impact is clearly visible and helps them prevent similar situations in the future. On our side, once the root cause became clear, we worked to automatically disable those packages from updating while the situation was being sorted out and to share recovery guidance as quickly as possible, and we genuinely appreciate the reports and troubleshooting details users shared along the way.

    Unfortunately, this wasn’t something we could safely automate. Systems were left in different states, with varying data layouts, database sizes, and available disk space, and recovery steps involved potentially destructive or irreversible actions. Automating that kind of remediation would have carried a real risk of data loss, especially for larger installations.

    If you’re still experiencing issues after following the recommended steps, please let us know — we’re happy to help take a closer look.

    1
  • Julien Venante

    Dan Muey - followed steps in fix. 9.7 was not initiated for me. Still unable to restart sqld.
    my.cnf keeps getting reverted with new variables and crashes the service.
    It reverts authentication_policy and mysql_native_password variables after commeting them out and trying a restart.

    0
  • Dan Muey

    Interesting Julien Venante I’ve not seen that personally. Feel free to open a ticket if you’d like us to have a closer look 👍

    0
  • Julien Venante

    Dan Muey, would love to but my licence is with OVH and I cannot open a ticket directly.
    Opening a ticket with them will be answered in a few years.

    0
  • Julien Venante

    I did a fatrace to see what's rewriting the file everytime and it's restartsrv_mysqld
    10:07:25.255539 restartsrv_mysq(60006): CW  /etc/my.cnf

    0
  • Julien Venante

    …Unrecognized configuration options may have caused the MySQL startup errors.
    The system has attempted to auto-update your MySQL configuration file for your MySQL version.

    0
  • Julien Venante

    [InnoDB] Invalid MySQL server downgrade: Cannot downgrade from 90700 to 80409. Downgrade is only permitted between patch releases.

    0
  • Damon Wright

    We also tried the fix and failed. Opened a ticket and AI triage suggested what we'd already documented we'd done, now waiting for human triage but feeling very frustrated! I wonder how many final straws we'll have with WebPros before we have broken backed camel.

    0
  • Damon Wright

    We have this too "[InnoDB] Invalid MySQL server downgrade: Cannot downgrade from 90700 to 80409. Downgrade is only permitted between patch releases."

    Or both initially and after reinstalling 9.7 this:
    "2026-04-23T15:21:57.776235Z 0 [ERROR] [MY-000067] [Server] unknown variable 'mysql_native_password=1'.
    2026-04-23T15:21:57.776722Z 0 [ERROR] [MY-010119] [Server] Aborting
    2026-04-23T15:21:59.325487Z 0 [System] [MY-010910] [Server] /usr/sbin/mysqld: Shutdown complete (mysqld 9.7.0)  MySQL Community Server - GPL."

    We've obviously tried commenting out the recommended variables. They either get added back or it attempts to rollback the configuration on restart. Neither 8.4.9 nor 9.7.0 will start.

    0
  • Julien Venante

    Damon Wright - we were able to fix it. I reinstalled 9.7 after downgrade, followed initiated fix procedure and finally was able to dump DBs and install 8.4. (then restore)

    0
  • MEPlus

    Thankfully, the downgrade procedure worked for me.  Not something I had on my agenda when I got up this morning, finding my mailbox filled with "mysql is DOWN" emails since 2 am this night.  Really not happy with this incident.

    0
  • cPRex Jurassic Moderator

    Update - April 23 - The details in the support article are correct, and it also appears that MySQL has resolved the upstream issue with their repositories. That doesn't help you if the damage has already happened to your server, but at this point we have a few options:

    -we've released an autofixer to help prevent users from receiving the update, which went out last night
    -the support article has copy/pastable steps to resolve the issue if your server experienced the upgrade
    -if you have *any* doubts at all, feel free to submit a ticket.

    0
  • Damon Wright

    Julien Venante Ditto. We reinstalled 9.7 and then followed the other set. It missed some 9.7 packages but they were easily manually removed. Thanks for thinking of us! The notification that someone had mentioned me on here dropped in just as we tested our first site was working correctly. 

    0
  • User404

    Seems like the upgrade broke our WHM/cPanel server as well. Upgrading happened 8 hours ago. So it's wrong information that this issue was resolved or any autofix did not help us. MySQL repos fixes did not save us eather.

    I had to restore MySQL 9.7 databases back to 8.4 via dump. As the workaround instructions said.

    After MySQL 8.4 was back on the system and databases are back in business: RoundCube Webmail stopped working. So there's that. RoundCube database is there, but mysql-user is missing. I hope nothing else is broke at the moment.

    I would like to know how to prevent this happening to other WHM servers?

    0
  • cPRex Jurassic Moderator

    User404 - performing the MySQL version lock is the best way to keep this from happening on other systems.  If you're seeing a server that *was* fixed on MySQL 8.4 and then it broke again after the nightly update, we'd like to take a look through a ticket.

    0
  • Amanpreet Bhangoo

    Did the "If the 9.7 init process failed, revert to MySQL 8.4 while MySQL 9 is offline." option. It's showing

    cPRex

    MySQL 8.4 is installed, but it fails to restart.

    Get the error:

    Waiting for “mysql” to start ……info [restartsrv_mysql] systemd failed to start the service “mysqld” (The “/usr/bin/systemctl restart mysqld.service --no-ask-password” command (process 106826) reported error number 1 when it ended.): Job for mysqld.service failed because the control process exited with error code.
    See "systemctl status mysqld.service" and "journalctl -xe" for details.

    1
  • cPRex Jurassic Moderator

    Can you create a ticket?  I really can't advise how to fix specific issues for these over the forum.

    0

Please sign in to leave a comment.