Difference between revisions of "ZFS administration"
From GridPP Wiki
(Created page with "==check pool status== <nowiki> [root@pool5 ~]# zpool status pool: tank-2TB state: ONLINE scan: scrub in progress since Mon May 16 11:30:03 2016 1.65T ...") |
m |
||
| Line 72: | Line 72: | ||
* unit is "per second" | * unit is "per second" | ||
* can be combined with per disk statistics: ''zpool iostat -v tank-2TB 1 5'' | * can be combined with per disk statistics: ''zpool iostat -v tank-2TB 1 5'' | ||
| + | |||
| + | ==Sharing via NFS== | ||
| + | Prerequisites on CentOS7. Start the services if you're not rebooting. | ||
| + | <nowiki> | ||
| + | yum install -y nfs-utils | ||
| + | systemctl enable zfs.target | ||
| + | systemctl enable zfs-import-cache.service | ||
| + | systemctl enable zfs-mount.service | ||
| + | systemctl enable zfs-share.service | ||
| + | systemctl enable nfs-server | ||
| + | firewall-cmd --permanent --zone=work --add-service=nfs | ||
| + | firewall-cmd --reload | ||
| + | </nowiki> | ||
| + | |||
| + | To share via NFS use the 'zfs set' command with a list of hosts/networks. This injects the details into the NFS exports table. | ||
| + | <nowiki> | ||
| + | zfs create tank/data | ||
| + | chmod 1777 /tank/data | ||
| + | zfs set sharenfs="rw=@physics*.example.ac.uk,rw=@foobar.example.ac.uk,rw=@172.16.0.0/24" tank/data | ||
| + | </nowiki> | ||
| + | |||
| + | Should now be mountable on the client. | ||
| + | <nowiki> | ||
| + | mount zfsserver:/tank/data /mnt | ||
| + | </nowiki> | ||
==Other helpful commands== | ==Other helpful commands== | ||
Revision as of 15:43, 22 February 2017
check pool status
[root@pool5 ~]# zpool status
pool: tank-2TB
state: ONLINE
scan: scrub in progress since Mon May 16 11:30:03 2016
1.65T scanned out of 59.4T at 125M/s, 134h57m to go
0 repaired, 2.79% done
config:
NAME STATE READ WRITE CKSUM
tank-2TB ONLINE 0 0 0
raidz3-0 ONLINE 0 0 0
scsi-36a4badb044e936001e55b25520b22e10 ONLINE 0 0 0
scsi-36a4badb044e936001e55b2111ca79173 ONLINE 0 0 0
scsi-36a4badb044e936001e55b26e2232640f ONLINE 0 0 0
scsi-36a4badb044e936001e55b2622171ff0b ONLINE 0 0 0
scsi-36a4badb044e936001e55b293245ae11b ONLINE 0 0 0
scsi-36a4badb044e936001e55b2d128122551 ONLINE 0 0 0
scsi-36a4badb044e936001e55b2461fc3cb38 ONLINE 0 0 0
scsi-36a4badb044e936001e55b3142c0ac749 ONLINE 0 0 0
scsi-36a4badb044e936001e55b27d230ce0f6 ONLINE 0 0 0
scsi-36a4badb044e936001e55b2b426603fbe ONLINE 0 0 0
scsi-36a4badb044e936001e55b2f42a2e3006 ONLINE 0 0 0
scsi-36a4badb044e936001e55b2c4274ec795 ONLINE 0 0 0
scsi-36a4badb044e936001e55b3082b59c1da ONLINE 0 0 0
scsi-36a4badb044e936001e55b31f2cbeb648 ONLINE 0 0 0
scsi-36a4badb044e936001e55b33b2e6172a4 ONLINE 0 0 0
scsi-36a4badb044e936001e55b34c2f70184c ONLINE 0 0 0
scsi-36a4badb044e936001e55b3793218970b ONLINE 0 0 0
scsi-36a4badb044e936001e55b36530e4cb2a ONLINE 0 0 0
scsi-36a4badb044e936001e55b39033741ccf ONLINE 0 0 0
scsi-36a4badb044e936001e830e00b16015f4 ONLINE 0 0 0
scsi-36a4badb044e936001e55b38532cf8c68 ONLINE 0 0 0
scsi-36a4badb044e936001e55b3f839a79c83 ONLINE 0 0 0
scsi-36a4badb044e936001e55b4023a46ae2a ONLINE 0 0 0
scsi-36a4badb044e936001e55b3eb38df509e ONLINE 0 0 0
scsi-36a4badb044e936001e55b42b3cb239cd ONLINE 0 0 0
scsi-36a4badb044e936001e55b41d3bdd5a86 ONLINE 0 0 0
scsi-36a4badb044e936001e55b40f3b0100cb ONLINE 0 0 0
scsi-36a4badb044e936001e55b4363d55d784 ONLINE 0 0 0
scsi-36a4badb044e936001e55b3de3822569a ONLINE 0 0 0
scsi-36a4badb044e936001e55b4413e09a9f1 ONLINE 0 0 0
scsi-36a4badb044e936001e55b44e3ecc77ea ONLINE 0 0 0
scsi-36a4badb044e936001e55b358301ee6a2 ONLINE 0 0 0
scsi-36a4badb044e936001e830de6afdb1d8f ONLINE 0 0 0
scsi-36a4badb044e936001e830df1b0805a46 ONLINE 0 0 0
scsi-36a4badb044e936001e830dfab1049dc8 ONLINE 0 0 0
errors: No known data errors
- Read: amount of read errors
- Write: amount of write errors
- Checksum: amount of wrongly read data from disk (checksum and data do not match)
- number usually only increases during scrub, but not when occured during normal operation
checking read/write/IOPS
[root@pool5 ~]# zpool iostat 1 5
capacity operations bandwidth
pool alloc free read write read write
---------- ----- ----- ----- ----- ----- -----
tank-2TB 59.4T 4.13T 407 656 47.9M 37.2M
tank-2TB 59.4T 4.13T 1.42K 0 167M 0
tank-2TB 59.4T 4.13T 1.54K 0 181M 0
tank-2TB 59.4T 4.13T 1.55K 0 182M 0
tank-2TB 59.4T 4.13T 1.45K 0 173M 0
- first line is average since switched on or imported
- can be suppressed with option "-y"
- unit is "per second"
- can be combined with per disk statistics: zpool iostat -v tank-2TB 1 5
Sharing via NFS
Prerequisites on CentOS7. Start the services if you're not rebooting.
yum install -y nfs-utils systemctl enable zfs.target systemctl enable zfs-import-cache.service systemctl enable zfs-mount.service systemctl enable zfs-share.service systemctl enable nfs-server firewall-cmd --permanent --zone=work --add-service=nfs firewall-cmd --reload
To share via NFS use the 'zfs set' command with a list of hosts/networks. This injects the details into the NFS exports table.
zfs create tank/data chmod 1777 /tank/data zfs set sharenfs="rw=@physics*.example.ac.uk,rw=@foobar.example.ac.uk,rw=@172.16.0.0/24" tank/data
Should now be mountable on the client.
mount zfsserver:/tank/data /mnt
Other helpful commands
- list all ZFS: zfs list
- list all zpools: zpool list
- history of all commands: zpool history
- list all events happened: zpool events
- scan the whole zpool for bad data: zpool scrub tank-2TB
- every block of data, metadata, and ditto-block is compared against it's checksum
- very long and I/O intensive process
- but runs with low priority during normal production access
- best to have cronjob setup to have it done automatically once per month/2 month/...
- get compressratio: zfs get compressratio
- import a zpool: zpool import NAME
- import all available zpools: zpool import -a
- see current cache statistics: arcstat.py 1
- zfs and zpool have very good man pages
- man zpool
- man zfs
