Not the answer you need?
Register and ask your own question!

Cluster seems properly synced, but clustercheck script fails

clott_memberclicksclott_memberclicks EntrantInactive User Role Beginner
Hello,

I have a completely new install of Percona Cluster 5.6 on CentOS 6.6 along the lines of the setup in the following blog post:

https://www.percona.com/blog/2012/06...-with-haproxy/

There are three hosts: each has a MySQL node, plus an HAProxy instance using the standard "clustercheck" script to health-check the local nodes via xinetd. All three hosts share a floating ip address via keepalived.

Every part of the stack except the clustercheck script seems to be working properly.

I.e, I get output like the following on all three hosts:
$ mysql -e "SHOW GLOBAL STATUS LIKE 'wsrep_cluster_%'"
+--------------------------+--------------------------------------+
| Variable_name | Value |
+--------------------------+--------------------------------------+
| wsrep_cluster_conf_id | 9 |
| wsrep_cluster_size | 3 |
| wsrep_cluster_state_uuid | dca6d39f-42dc-11e5-946f-b7d45ebb1bba |
| wsrep_cluster_status | Primary |
+--------------------------+--------------------------------------+

I also created users on one of the nodes, and I can see that it propagated as it should to mysql.user on all cluster members:
mysql> SELECT Host, User FROM mysql.user;
+----------------------+----------+
| Host | User |
+----------------------+----------+
| % | agunn |
| % | ahough |
| % | bharden |
| % | clott |
| % | gloy |
| % | joomla |
| % | mcdba |
| 127.0.0.1 | root |
| ::1 | root |
| localhost | mysqlchk |
| localhost | root |
| localhost | sstuser |
+----------------------+----------+
12 rows in set (0.00 sec)

In HAProxy, all three nodes are failing the health check via xinetd.

I know the xinetd configuration is working because I can reproduce the failed health checks via curl:
$ curl [URL]http://joomdb01:9200/[/URL]
Percona XtraDB Cluster Node is not synced or non-PRIM.

And when I run the clustercheck script manually at the command line outside of xinted, it returns the same output:
$ /usr/bin/clustercheck [actual_password_redacted] 1 /var/log/mysqlchk 0 /etc/my.cnf
HTTP/1.1 503 Service Unavailable
Content-Type: text/plain
Connection: close
Content-Length: 57

Percona XtraDB Cluster Node is not synced or non-PRIM.

Basically, it looks like everything except the clustercheck script is working as intended. Any ideas as to where I should be looking?

Thanks,

-Cal

Comments

  • clott_memberclicksclott_memberclicks Entrant Inactive User Role Beginner
    So it turns out this was a typo in the clustercheck user name. I corrected the provisioned userid in MySQL and all was well.

    The clustercheck script does not have very informative error message for this particular failure scenario.
Sign In or Register to comment.

MySQL, InnoDB, MariaDB and MongoDB are trademarks of their respective owners.
Copyright ©2005 - 2020 Percona LLC. All rights reserved.