mirror of
https://github.com/percona/percona-toolkit.git
synced 2025-09-11 05:29:30 +00:00
76a010abee
This turned out to be two bugs mangled into one.
First, _parse_varvals can deal with (var, undef), but not with (undef).
This is a problem because two of the trhee spots that call
_parse_varvals can return undef because of this:
map { $_ =~ m/^([^=]+)(?:=(.*))?$/ }
grep { $_ !~ m/^\s*#/ } # no # comment lines
split("\n", $mysqld_section)
The problem is twofold. First, we are not skipping empty or
whitespace-only lines. That means that the map will fail,
and pass an undef to _parse_varvals. So this ended up in
a triple fix: Make _parse_varvals deal with a sole undef,
skip empty/whitespace lines, and change that map to
map { $_ =~ m/^([^=]+)(?:=(.*))?$/ ? ($1, $2) : () }
so even if the regex fails in the future, no sole undef
will be passed down the chain.
31 lines
1.0 KiB
INI
31 lines
1.0 KiB
INI
[client]
|
|
user = msandbox
|
|
password = msandbox
|
|
port = PORT
|
|
socket = /tmp/PORT/mysql_sandboxPORT.sock
|
|
|
|
[mysqld]
|
|
port = PORTa
|
|
socket = /tmp/PORT/mysql_sandboxPORT.sock
|
|
pid-file = /tmp/PORT/data/mysql_sandboxPORT.pid
|
|
basedir = PERCONA_TOOLKIT_SANDBOX
|
|
datadir = /tmp/PORT/data
|
|
key_buffer_size = 16M
|
|
innodb_buffer_pool_size = 16M
|
|
innodb_data_home_dir = /tmp/PORT/data
|
|
innodb_log_group_home_dir = /tmp/PORT/data
|
|
innodb_data_file_path = ibdata1:10M:autoextend
|
|
innodb_log_file_size = 5M
|
|
log-bin = mysql-bin
|
|
relay_log = mysql-relay-bin
|
|
log_slave_updates
|
|
server-id = PORT
|
|
report-host = 127.0.0.1
|
|
report-port = PORT
|
|
log-error = /tmp/PORT/data/mysqld.log
|
|
innodb_lock_wait_timeout = 3
|
|
general_log
|
|
general_log_file = genlog
|
|
# These next two lines have a space in them, and trigger the bug
|
|
|
|
|