Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Fixed
-
10.0.14-galera
-
CentOS 6.5
Description
When using binlog_checksum = 'CRC32', galera fails to replicate DDL events. I don't currently have an environment where I can test this more so I'm going off binlogs and error logs from when I had binlog_checksum set and not.
When binlog_checksum was set, I saw errors like in mysqld_checksum_errors.txt
. The errors on the analyze queries aren't critical but the error when creating the table caused two of my three nodes to kill themselves since the originating node then proceeded to load data into the table. Having set binlog_checksum to 'NONE' (the default) my cluster handles DDL queries normally now. I looked in the galera cache file on the surviving node and the queries did have the extra bytes after them just as in the error log.
I also attached the GRA log files from the queries mentioned in the log file.
Attachments
Activity
| Field | Original Value | New Value |
|---|---|---|
| Attachment | my.cnf [ 36304 ] |
| Attachment | GRA_96_336539821.log [ 36305 ] | |
| Attachment | GRA_66_335743799.log [ 36306 ] | |
| Attachment | GRA_41_335743798.log [ 36307 ] | |
| Description |
When using binlog_checksum = 'CRC32', galera fails to replicate DDL events. I don't currently have an environment where I can test this more so I'm going off binlogs and error logs from when I had binlog_checksum set and not. When binlog_checksum was set, I saw errors like in [^mysqld_checksum_errors.txt]. The errors on the analyze queries aren't critical but the error when creating the table caused two of my three nodes to kill themselves since the originating node then proceeded to load data into the table. Having set binlog_checksum to 'NONE' (the default) my cluster handles DDL queries normally now. I looked in the galera cache file on the surviving node and the queries did have the extra bytes after them just as in the error log. |
When using binlog_checksum = 'CRC32', galera fails to replicate DDL events. I don't currently have an environment where I can test this more so I'm going off binlogs and error logs from when I had binlog_checksum set and not. When binlog_checksum was set, I saw errors like in [^mysqld_checksum_errors.txt]. The errors on the analyze queries aren't critical but the error when creating the table caused two of my three nodes to kill themselves since the originating node then proceeded to load data into the table. Having set binlog_checksum to 'NONE' (the default) my cluster handles DDL queries normally now. I looked in the galera cache file on the surviving node and the queries did have the extra bytes after them just as in the error log. I also attached the GRA log files from the queries mentioned in the log file. |
| Assignee | Nirbhay Choubey [ nirbhay_c ] | |
| Labels | galera |
| Status | Open [ 1 ] | In Progress [ 3 ] |
| Fix Version/s | 5.5.43-galera [ 18819 ] | |
| Fix Version/s | 10.0.18-galera [ 18820 ] | |
| Resolution | Fixed [ 1 ] | |
| Status | In Progress [ 3 ] | Closed [ 6 ] |
| Workflow | MariaDB v2 [ 58712 ] | MariaDB v3 [ 67161 ] |
| Workflow | MariaDB v3 [ 67161 ] | MariaDB v4 [ 148539 ] |
https://github.com/MariaDB/server/commit/d7445ea6dfbd3be390792c636f755d3bb40d5fbf