Details
-
Bug
-
Status: Closed (View Workflow)
-
Minor
-
Resolution: Fixed
-
10.0.12-galera
-
CentOS 6.5 VM running on Linode environment
Description
Ran a yum upgrade, and following this yum hung. Opened a second shell, and found the attached error in the Mysql Log. Tried to restart with identical result: instant crash. Other nodes don't even see attempt to re-connect. Only way to fix it was to move my /var/lib/mysql directory and redo the SST.
Please let me know if I can provide any further info, I've tried to preserve as much as possible to help out with this.
Attachments
Activity
| Field | Original Value | New Value |
|---|---|---|
| Assignee | Nirbhay Choubey [ nirbhay_c ] |
Oops, apologies. This was the second crash, after I attempted to roll back. Attached the original one.
| Attachment | MariaDB-Crash-Orig.txt [ 32603 ] |
I couldn't reproduce it on CentoOS 6.5 32-bit, 10.0.11-galera -> 10.0.12-galera. Looking at both the attached files, it looks like crash occurred for both the versions during plugin initialization. Just to make myself clear : the node started fine after you removed the datadir for SST? Can you share the complete server error log and configurations?
Sorry for the delay here. The setup is a three node master-master mariadb cluster, all running CentOS 6.5. The process to break was:
1. yum update on one of the nodes
2. MariaDB crashed during yum update, freezing yum.
3. Manually killed yum, tried to restart mariadb, instant crash
4. Ran yum remove 10.0.12-galera, downloaded and re-installed 10.0.11
5. Tried to restart mariadb, instant crash
6. Moved grastate file into home directory, tried to restart mariadb, instant crash
7. Rsync'd entire /var/lib/mysql into home directory, rm -rf /var/lib/mysql/* and then restarted mariaDB, SST completed and runs fine.
I still have all the files. Have uploaded the entire mysql.log and copies of my mysql config.
| Attachment | mysql.log [ 32800 ] |
| Attachment | cluster.cnf [ 32801 ] |
| Status | Open [ 1 ] | In Progress [ 3 ] |
Hi there,
Yes I do, and my entire cluster now refuses to start. I can start them individually, but as soon as I try and enable clustering it crashes in fiery glory. Understandably, I'm hesitant to upload my data directory to the internet, but I'm willing to share it privately with you to help try and pin this down. Can you email me? mariaDB AT falconsupport.com.
Thanks!
Andy
Can you upload them to ftp.askmonty.org?
https://mariadb.com/kb/en/mariadb/development/debugging-mariadb/reporting-bugs/
I'd prefer to remove the sensitive information first... which rather removes the point of me uploading it I guess. What are you looking for in particular, the data files or the database contents?
I see. What is the SE for mysql.plugin table? MyISAM or InnoDB?
SHOW CREATE TABLE mysql.plugin;
| Assignee | Nirbhay Choubey [ nirbhay_c ] | Jan Lindström [ jplindst ] |
| Status | In Progress [ 3 ] | In Review [ 10002 ] |
| Assignee | Jan Lindström [ jplindst ] | Nirbhay Choubey [ nirbhay_c ] |
| Status | In Review [ 10002 ] | Stalled [ 10000 ] |
Pushed to maria-10.0-galera.
http://bazaar.launchpad.net/~maria-captains/maria/maria-10.0-galera/revision/3899
| Component/s | Galera [ 10124 ] | |
| Fix Version/s | 10.0.14-galera [ 17000 ] | |
| Resolution | Fixed [ 1 ] | |
| Status | Stalled [ 10000 ] | Closed [ 6 ] |
| Workflow | MariaDB v2 [ 50709 ] | MariaDB v3 [ 64896 ] |
| Workflow | MariaDB v3 [ 64896 ] | MariaDB v4 [ 148059 ] |
For a note: the log says "Server version: 10.0.11-MariaDB-wsrep", so apparently upgrade didn't work well enough.