Add SYSTEMD_READWRITEPATH-variable to mariadb{@,}.service.in to make sure that
if one is not building RPM or DEB packages then make sure there is ReadWritePaths
directive is defined in systemd service file.
This ensures that tar-ball installation has permissions to write database default
installation path (default: /usr/local/mysql/data) even if it's located
under /usr. Writing to that location is prevented by 'ProtectSystem=full'
systemd directive by default.
Prefixing the path with "-" in systemd causes there to not be an error if the
path doesn't exist. This may occur if the user has configured a datadir
elsewhere.
Reviewer: Daniel Black
Quoting MDEV reporter Daniel Lewart:
Starting MariaDB with default configuration causes the following problems:
"[Warning] Could not increase number of max_open_files to more than 16384 (request: 32186)"
silently reduces table_open_cache_instances from 8 (default) to 4
Default Server System Variables:
extra_max_connections = 1
max_connections = 151
table_open_cache = 2000
table_open_cache_instances = 8
thread_pool_size = 4
LimitNOFILE=16834 is in the following files:
support-files/mariadb.service.in
support-files/mariadb@.service.in
Looking at sql/mysqld.cc lines 3837-3917:
wanted_files= (extra_files + max_connections + extra_max_connections +
tc_size * 2 * tc_instances);
wanted_files+= threadpool_size;
Plugging in the default values:
wanted_files = (30 + 151 + 1 + 2000 * 2 * 8 + 4) = 32186
However, systemd configuration has LimitNOFILE = 16384, which is far smaller.
I suggest increasing LimitNOFILE to 32768.
Replace all references to /usr/sbin/mysqld (and bin and libexec) with
mariadbd, so that the binary server will always be 'mariadbd'.
Also update all places that reference the server binary in other ways,
such as AppArmor profiles and scripts that previously expected to find
a 'mysqld' in process lists.
When trying to start mariadb via systemctl, WSREP failed
to start mysqld for wsrep recovery, because the binary
"galera-recovery" is neither searching the mysqld in the
same folder as the binary itself nor in the path variable
but instead expects the root to be /usr/local/mysql.
This fix changes the current directory to the desired
directory before starting mysqld.
The arg was introduced as part of 75bcf1f9ad
to fix a SELinux problem caused by mysqld_safe accessing files it should
not be via the my_which function.
The root cause for this was fixed in 10.3, via
355ee6877b which eliminated the my_which
function from mysqld_safe entirely. Thus, in 10.3, this --basedir flag
is not necessary.
The unit files made systemd print:
systemd[1]: Started MariaDB 10.3.13 database server (multi-instance).
Let's add the instance name, so starting mariadb@foo.service
makes it print:
systemd[1]: Started MariaDB 10.3.13 database server (multi-instance foo).
Include comment header that describes overrides.
Unit description now includes @VERSION@.
After=syslog.target removed - redunant
Add --basedir=@prefix to prevent /root/.my.cnf lookups. This is
placed after $MYSQLD_OPTIONS in case a user sets a --{no,}default
type options which has to be first in the mysqld arguements.
Additional changes to multi instance (support-files/mariadb@.service.in):
* added @SYSTEMD_EXECSTARTPRE@ / @SYSTEMD_EXECSTARTPOST@
* removed mariadb@bootstrap reference as galera_new_cluster as
it's a little too proment.
* use_galera_new_cluster.conf updated to override pre/post steps
to ensure it has no side effects
Signed-off-by: Daniel Black <daniel@linux.vnet.ibm.com>
* wait() for the child process to die, let it rest in peace
* fix incorrect parentheses
* if there was no password on the command line or in .cnf file,
pkt will be "", and we need to request the user to enter the password
* make sure that auth->salt is always allocated on a permanent memroot.
when called from set_user_salt_if_needed(), user_copy and its auth_str
are on the thd memroot, but auth_copy->salt is then copied to auth->salt
* adjust service files so that systemd wouldn't interfere with our
setuid executables
also
* print the pam error message in debug mode
By removing Galera functionality, we remove PermissionsStartOnly=true
and hence make this service more flexible for running multiple
instances each on a different user.
When the multi-instance systemd service file was chosen it effectively
relied on /etc/my.cnf.d/my{instancename}.cnf file to define its
configuration file. This is problematic if running along side a
single instance mariadb service which has /etc/my.cnf that reads all
configuration file /etc/my.cnf.d/*.cnf.
To prevent the service from auto starting up if a user has this
previous configuration ConditionPathExists=!@sysconf2dir@/my%I.cnf
to ensure that a user with the previous configuration isn't
started in a non-intended mode. Documentation in the service file
(should be release notes too), described a recommended migration.
A new approach was to use --defaults-group-suffix=.%I as an
arguement to mysqld and let the user define a [mysqld.{instancename}]
group within the configuration file. This way existing global
mysqld configuration options are read with the instance name
having special overrides of datadir, port, socket etc.
A systemd environment variable MYSQLD_MULTI_INSTANCE is used in the
defination as it give the user flexability to use multiple
segregation mechanisms between services. This is used multiple
times within the service which all needed to be kept consistent.
Another notable change is mysql_install_db being part of the
ExecStartPre. This provides and auto-initialization for users
that run multiple instances.
When galera is used we want a stop to kill off not only the mysqld
process but the entire process group created by galera to perform
sst or wsrep_notify_cmd.`
Galera recovery process works in two phases. In the first
phase, mysqld is started as non-daemon with --wsrep-recover
to recover and fetch the last logged global transaction ID.
This ID is then used in second phase as the start position
(--wsrep-start-position=XX) to start mysqld as daemon.
As this process was implemented in mysqld_safe script, the
recovery did not work when server was started using systemd.
Fixed by introducing a shell script (wsrep_recovery.sh) that
mimics the first phase of the recovery process.
--defaults-xxx options must be placed before all other
options in the command line. Also moved MYSQLD_OPTS at
the end so that its options take precedence.
