Munin-run thesql_ plugin returns data but munin server gets ‘Bad exit’ error

I've a Munin server with several Munin nodes, all configured using Ansible so configuration it's pretty the same. Systems are Debian Jessie.

The mysql_* graphs from two of the servers are present but empty while the other three servers have full graphs.

Running munin-run --debug mysql_commands in one of the faulty servers returns correct output:

# munin-run --debug mysql_commands
# Processing plugin configuration from /etc/munin/plugin-conf.d/ansible.conf
# Processing plugin configuration from /etc/munin/plugin-conf.d/munin-node
# Setting /rgid/ruid/ to /119/0/
# Setting /egid/euid/ to /119 119/0/
# Setting up environment
# Environment mysqluser = debian-sys-maint
# Environment mysqlconnection = DBI:mysql:mysql;mysql_read_default_file=/etc/mysql/debian.cnf
# Environment mysqlopts = --defaults-file=/etc/mysql/debian.cnf
# About to run '/etc/munin/plugins/mysql_commands'
Com_delete.value 4546376
Com_insert.value 2804559
Com_insert_select.value 341479
Com_load.value 0
Com_replace.value 0
Com_replace_select.value 0
Com_select.value 236967004
Com_update.value 7069348
Com_update_multi.value 0

So it seems Munin node is working ok.

But when I execute munin-cron in the munin server the plugin returnsbad exit when fetched:

# sudo -u munin munin-cron --host server.com --debug
...
2018/01/09 08:35:01 [DEBUG] for my mysql_commands (irqstats mysql_innodb_tnx diskstats apache_accesses smart_sda df ntp_offset vmstat fw_packets mysql_select_types netstat users mysql_qcache mysql_table_locks swap mysql_slow mysql_myisam_indexes if_err_eth0 apt hddtemp_smartctl ntp_kernel_err ntp_kernel_pll_freq mysql_innodb_rows mysql_innodb_bpool forks ntp_kernel_pll_off cpu load postfix_mailvolume postfix_mailqueue mysql_qcache_mem open_inodes processes http_loadtime open_files uptime mysql_commands mysql_innodb_semaphores apt_all mysql_innodb_io smart_sdb memory mysql_connections mysql_files_tables mysql_network_traffic df_inode mysql_sorts if_eth0 entropy proc_pri mysql_innodb_bpool_act apache_processes threads mysql_innodb_log mysql_innodb_insert_buf mysql_tmp_tables mysql_innodb_io_pend interrupts apache_volume)
2018/01/09 08:35:01 [DEBUG] Fetching service configuration for 'mysql_commands'
2018/01/09 08:35:01 [DEBUG] Writing to socket: "config mysql_commands
".
2018/01/09 08:35:02 [DEBUG] Reading from socket: "graph_vlabel Commands per ${graph_period}\ngraph_total Questions\ngraph_args --base 1000\ngraph_title Command Counters\ngraph_category mysql2\nCom_delete.type DERIVE\nCom_delete.label Delete\nCom_delete.draw STACK\nCom_delete.min 0\nCom_insert.draw STACK\nCom_insert.min 0\nCom_insert.type DERIVE\nCom_insert.label Insert\nCom_insert_select.type DERIVE\nCom_insert_select.label Insert select\nCom_insert_select.draw STACK\nCom_insert_select.min 0\nCom_load.label Load Data\nCom_load.type DERIVE\nCom_load.min 0\nCom_load.draw STACK\nCom_replace.type DERIVE\nCom_replace.label Replace\nCom_replace.draw STACK\nCom_replace.min 0\nCom_replace_select.draw STACK\nCom_replace_select.min 0\nCom_replace_select.type DERIVE\nCom_replace_select.label Replace select\nCom_select.type DERIVE\nCom_select.label Select\nCom_select.draw STACK\nCom_select.min 0\nCom_update.type DERIVE\nCom_update.label Update\nCom_update.draw STACK\nCom_update.min 0\nCom_update_multi.label Update multi\nCom_update_multi.type DERIVE\nCom_update_multi.min 0\nCom_update_multi.draw STACK".
2018/01/09 08:35:02 [DEBUG] config: 0.151266 sec for 'mysql_commands' on server.com/1.2.3.4/4949
2018/01/09 08:35:02 [DEBUG] Now parsing config output from plugin mysql_commands on server.com
2018/01/09 08:35:02 [DEBUG] update_rate 0 for mysql_commands on server.com/1.2.3.4:4949
2018/01/09 08:35:02 [DEBUG] No service data for mysql_commands, fetching it
2018/01/09 08:35:02 [DEBUG] Writing to socket: "fetch mysql_commands
".
2018/01/09 08:35:02 [DEBUG] data: 0.175058 sec for 'mysql_commands' on server.com/1.2.3.4/4949
2018/01/09 08:35:02 [DEBUG] Now parsing fetch output from plugin mysql_commands on server.com/1.2.3.4:4949
2018/01/09 08:35:02 [FETCH from mysql_commands] # Bad exit
...

This happens with all mysql_* plugins in two servers, but they work ok in other three servers. Other Munin plugin work ok in all servers, so general configuration seems ok.

Configuration for mysql_* plugins:

[mysql*]
user root
env.mysqlopts --defaults-file=/etc/mysql/debian.cnf
env.mysqluser debian-sys-maint
env.mysqlconnection DBI:mysql:mysql;mysql_read_default_file=/etc/mysql/debian.cnf

No new entries are added to Munin node log (/var/log/munin/munin-node.log) nor MySQL when sudo -u munin munin-cron is executed.

Munin node service reports normal operation:

# systemctl status munin-node
● munin-node.service - Munin Node
   Loaded: loaded (/lib/systemd/system/munin-node.service; enabled)
   Active: active (running) since mié 2018-01-17 14:20:10 CET; 19s ago
     Docs: man:munin-node(1)
           http://munin.readthedocs.org/en/stable-2.0/reference/munin-node.html
  Process: 4515 ExecStart=/usr/sbin/munin-node $DAEMON_ARGS (code=exited, status=0/SUCCESS)
 Main PID: 4541 (munin-node)
   CGroup: /system.slice/munin-node.service
           └─4541 /usr/bin/perl -wT /usr/sbin/munin-node

ene 17 14:20:10 example.com systemd[1]: Started Munin Node.

• Munin version (nodes and servers): Debian package 2.0.25-1+deb8u3.

• Database: mariadb-server Debian Package 10.0.26-0+deb8u1.

What can be wrong with those nodes? Or how can I debug the problem?