mariadb fails to start
depasseg opened this issue · 5 comments
On a centos 7.4 system, the /roles/mariadb/tasks/main.yml fails when trying to start mariadb after successfully creating the configuration file, the log file and PID directory.
A manual yum install of mariadb-server works fine.
Logs attached:
logs.txt
job_35.txt
I used this to install mariadb to that host, and then cloudera installation continued to run.
a buffer of innodb_buffer_pool_size=... has a size comparable to host memory?
@depasseg - Please check if selinux is disabled, I was able to resolve this issue after disabling selinux.
I was able to resolve this issue after disabling selinux.
Same for me, works after setting it permissive ( setenforce 0).
A check in the audit shows that somehow there seems a permission problem with the /var/log/mysqld.log:
grep mysql /var/log/audit/audit.log
Same info here:https://dba.stackexchange.com/questions/86129/could-not-start-mysql-or-mariadb-permissions-were-ok-but-mysql-bin-index-permi#136591
Still wondering why exactly?!
I used this to install mariadb to that host, and then cloudera installation continued to run.
That ansible-role-mariadb role looks like a good choice , allowing fully customized config.
The mariadb deployment (builtin cloudera role) writes a lot of the configs hardcoded .. and is IMO missing >1 important setting, so the mysqld listens on all interfaces:
bind-address=*
More infos, 1.cfg option shown in https://github.com/bertvv/ansible-role-mariadb#basic-configuration