GFS2 failisüsteemi kasutamine
Sisukord
Sissejuhatus
GFS2 (Global File System) http://sources.redhat.com/cluster/gfs/ on ... (ingl. k. shared disk cluster file system).
Tööpõhimõte
TODO
Tarkvara paigaldamine debianis
# apt-get install gfs2-tools
Tarkvara paigaldamine centosis
# yum groupinstall "iSCSI Storage Client" "High Availability" "Resilient Storage"
Seadistamine
/etc/hosts faili defineerime kõik workerid
10.100.0.1 moodle1 10.100.0.2 moodle2 10.100.0.3 moodle3
cmani seadistamine
et clvmd ja cman töötaks, on vaja /etc/cluster/cluster.conf tekitada. Selleks igas nodes käsud
ccs -f /etc/cluster/cluster.conf --createcluster moodle ccs -f /etc/cluster/cluster.conf --addnode moodle1 ccs -f /etc/cluster/cluster.conf --addnode moodle2 ccs -f /etc/cluster/cluster.conf --addnode moodle3
Testime konfi
# ccs_config_validate Configuration validates
et cman ka ilma quorumita starditaks, muidu ta hakkab kurjustama ja keeldub käivitumast ja tekib nokk kinni ja saba lahti stiilis ring
# echo "CMAN_QUORUM_TIMEOUT=0" >> /etc/sysconfig/cman
Teenused käima, centosis sedasi
# service cman start
Või debianis
# /etc/init.d/cman start Starting cluster: Checking Network Manager... [ OK ] Global setup... [ OK ] Loading kernel modules... [ OK ] Mounting configfs... [ OK ] Starting cman... [ OK ] Waiting for quorum... [ OK ] Starting fenced... [ OK ] Starting dlm_controld... [ OK ] Starting gfs_controld... [ OK ] Unfencing self... [ OK ] Joining fence domain... [ OK ]
Vaatame kas quorum on koos ehk kõik masinad suhtlevad
# corosync-quorumtool -l Nodeid Name 1 moodle1 2 moodle2 3 moodle3
kus
- cman two_node="1" expected_votes="1" - vajalik 2 komponendilise DRBD süsteemi puhul juhutumiks kui üks kaob
Clmvmd ja lvm tööle
lvm-i konfi muuta /etc/lvm/lvm.conf
locking_type = 3 fallback_to_local_locking = 0
Lvm tekitada, seadmeks meil multi0 mis tekitatud iscsi+multipathiga
pvcreate /dev/mapper/multi0 vgcreate clustervg /dev/mapper/multi0 vgchange -cy clustervg lvcreate -l 100%FREE -n moodledata clustervg lvmconf --enable-cluster
Ja samuti teenused tööle
# service clvmd start
GFS tekitamine
# mkfs.gfs2 -p lock_dlm -t moodle:moodledata -j 2 /dev/clustervg/moodledata This will destroy any data on /dev/drbd1. It appears to contain: LVM2 (Linux Logical Volume Manager) , UUID: THHRQpAS11PRt2eImKtkh7pxZOiOa3U Are you sure you want to proceed? [y/n] y Device: /dev/drbd1 Blocksize: 4096 Device Size 100.00 GB (26213591 blocks) Filesystem Size: 100.00 GB (26213591 blocks) Journals: 2 Resource Groups: 400 Locking Protocol: "lock_dlm" Lock Table: "network-raid1:*" UUID: C920F2FB-6E5D-57B9-C1AD-964FE8FDA3E0
kusjuures -t väärtused peavad kattuma cluster.conf failis oleva cluster nimega ja resource nimega
# gfs2_tool gettune /srv/gfs2 incore_log_blocks = 1024 log_flush_secs = 60 quota_warn_period = 10 quota_quantum = 60 max_readahead = 262144 complain_secs = 10 statfs_slow = 0 quota_simul_sync = 64 stall_secs = 600 statfs_quantum = 30 quota_scale = 1.0000 (1, 1) new_files_jdata = 0
Kui kolmas node annab käivitamisel veate
Mounting GFS2 filesystem (/GFS): Too many nodes mounting filesystem, no free journals
Lisame täiendava journali
# gfs2_jadd -j 1 /dev/mapper/clustervg-moodledata Filesystem: /GFS Old Journals 2 New Journals 3
Vaatame gfs2 enda staatust
# clustat Cluster Status for moodle @ Wed May 18 15:20:02 2016 Member Status: Quorate Member Name ID Status ------ ---- ---- ------ moodle1 1 Online moodle2 2 Online moodle3 3 Online, Local
Kui väga hädasti veebiliidest ka vaja siis pigaldada luci
/etc/init.d/luci restart Point your web browser to https://moodle1:8084 (or equivalent) to access luci
Kasulikud lisamaterjalid
- http://www.redhat.com/docs/en-US/Red_Hat_Enterprise_Linux/5/html/Global_File_System_2/ch-overview-GFS2.html
- http://gcharriere.com/blog/?p=73
- http://wiki.debian.org/kristian_jerpetjoen
- http://gfs.wikidev.net/DRBD_Cookbook
- http://sourceware.org/cluster/wiki/DRBD_Cookbook
- http://en.wikipedia.org/wiki/Global_File_System
- http://sources.redhat.com/cluster/doc/usage.txt
- http://sources.redhat.com/cluster/wiki/CategoryHowTo
