FreeBSD Cluster Active/Passive MySQL (iscsi, carp, devd, mysql)

Ciao, volevo condividere qualche spunto di riflessione per un cluster
Active/Passive MySQL con partizione dei dati e binlog su share iscsi
utilizzando carp e devd.

Sembra che utilizzare centos/corosync/pacemaker abbia una letteratura più
copiosa, e passi per essere un sistema più robusto, ma - onestamente - non
mi piace molto un sistema (pacemaker) che da una release (1.1.7) all'altra
(1.1.8) cambi l'utility principale (crm -> pcs) e la relativa sintassi dei
comandi.

Voi avete qualche setup del genere? Avete già testato qualcosa? Avete
spunti/consigli/desiderata?

Di seguito il mio setup di test.

Nodi:
- fbsd10-1
- fbsd10-2

[ fbsd10-1]:/etc/iscsi.conf
------------------------------------------------------------------------------
disk1 {
    authmethod      = CHAP
    chapIName       = user1
    chapSecret      = secret123456
    initiatorname   = fbsd10-1
    TargetName      = iqn.2013-10.reversed.domain.name:disk1
    TargetAddress   = 192.168.17.26:3260,1
}
------------------------------------------------------------------------------

[ fbsd10-2]:/etc/iscsi.conf
------------------------------------------------------------------------------
disk1 {
    authmethod      = CHAP
    chapIName       = user1
    chapSecret      = secret123456
    initiatorname   = fbsd10-2
    TargetName      = iqn.2013-10.reversed.domain.name:disk1
    TargetAddress   = 192.168.17.26:3260,1
}
------------------------------------------------------------------------------

[ fbsd10-1]:/etc/rc.conf
...
ifconfig_em0="inet 192.168.20.61 netmask 255.255.248.0"
ifconfig_em0_alias0="inet 192.168.20.60 netmask 255.255.248.0 vhid 60 pass
r0cksol1d advbase 1"
...

[ fbsd10-2]:/etc/rc.conf
...
ifconfig_em0="inet 192.168.20.62 netmask 255.255.248.0"
ifconfig_em0_alias0="inet 192.168.20.60 netmask 255.255.248.0 vhid 60 pass
r0cksol1d advbase 1"
...

[ALL]:/etc/rc.conf.local
...
mysql_enable="NO"
mysql_dbdir="/DATA/mysql"

[ALL]:/etc/devd.conf
...
notify 0 {
match "system"          "CARP";
        match "subsystem"       "[0-9]+@[0-9a-z]+";
        match "type"            "MASTER";
        action "/usr/local/vbin/master.sh";
};

notify 0 {
match "system"          "CARP";
        match "subsystem"       "[0-9]+@[0-9a-z]+";
        match "type"            "BACKUP";
        action "/usr/local/vbin/slave.sh";
};

[ALL]:/usr/local/vbin/master.sh
#!/bin/sh

. /usr/local/vbin/vars

# Master node

sleep 5

# Grab iSCSI resource
/sbin/iscontrol -c /etc/iscsi.conf -p /var/run/iscontrol.pid -n disk1
if [ ! $? -eq 0 ]; then
  echo "ERROR: Problems attaching iSCSI resource." | logger -p local1.err
  exit 255
fi

sleep 5

if [ ! -e $ISCSIDEV ]; then
  echo "ERROR: Problem mounting iSCSI resource." | logger -p local1.err
  exit 255
fi

echo "INFO: iSCSI resource attached successfully." | logger -p local1.info

# Fsck DATA partition
/sbin/fsck_ufs -By $ISCSIDEV
if [ ! $? -eq 0 ]; then
  echo "ERROR: Problems fscking ${ISCSIDEV}." | logger -p local1.err
  exit 255
fi

echo "INFO: Fsck completed successfully." | logger -p local1.info

# Mounting DATA partition
/sbin/mount $ISCSIDEV $MOUNTPNT
if [ ! $? -eq 0 ]; then
  echo "ERROR: Problems mounting ${MOUNTPNT}." | logger -p local1.err
  exit 255
fi

echo "INFO: ${MOUNTPNT} mounted." | logger -p local1.info

# Starting MySQL
$MYSQLSERVICE onestart
if [ ! $? -eq 0 ]; then
  echo "ERROR: Problems starting MySQL." | logger -p local1.err
  exit 255
fi

echo "INFO: MySQL started successfully." | logger -p local1.info

exit 0

[ALL]:/usr/local/vbin/slave.sh
#!/bin/sh

. /usr/local/vbin/vars

# Slave node

sleep 5

# Stopping MySQL
/usr/local/etc/rc.d/mysql-server onestop
$MYSQLSERVICE onestop

echo "INFO: MySQL stopped." | logger -p local1.info

# Umounting DATA partition
umount $ISCSIDEV $MOUNTPNT
ISMOUNTED=$(mount | grep "^$ISCSIDEV" | wc -l)
if [ $ISMOUNTED -ne 0 ]; then
  echo "ERROR: Unable to umount $MOUNTPNT" | logger -p local1.err
  exit 255
fi

echo "INFO: ${MOUNTPNT} umounted successfully." | logger -p local1.info

# Releasing iSCSI resource
ISCPID=$(cat /var/run/iscontrol.pid)
kill -HUP $ISCPID

echo "INFO: iSCSI resource released successfully." | logger -p local1.info

exit 0

[ALL]:/usr/local/vbin/vars
#!/bin/sh

ISCSIDEV="/dev/da0s1a"
MOUNTPNT="/DATA"
MYSQLSERVICE="/usr/local/etc/rc.d/mysql-server"


All'inizio occorre montare manualmente la share iscsi su un nodo e
formattarla:
fdisk -I /dev/da0
bsdlabel -w /dev/da0s1
newfs -j /dev/da0s1a

A questo punto un bel reboot di entrambi i nodi e via.

Se avete syslog configurato correttamente, avrete i messaggi degli script
sulla facility local1.


-- 
d.

_______________________________________________
Esperti mailing list
[email protected]
http://mailman.gufi.org/mailman/listinfo/esperti