[lustre-discuss] quotas not being enforced
Andrew Elwell
andrew.elwell at gmail.com
Wed Jan 13 23:28:39 PST 2021
On Thu, 14 Jan 2021 at 17:12, Andrew Elwell <andrew.elwell at gmail.com> wrote:
> I'm struggling to debug quota enforcement (or more worryingly, lack
> of) in recentish LTS releases.
>
> [root at pgfs-mds3 ~]# lctl conf_param testfs.quota.ost=g
> ... time passes
> [root at pgfs-mds4 ~]# lctl get_param osd-*.*.quota_slave.info | egrep
> '(name|enabled)'
> target name: testfs-OST0000
> quota enabled: none
> target name: testfs-OST0001
> quota enabled: none
>
> doesn't seem to be rippling out.
Not sure I've done the right thing, but poking via
lctl set_param osd-ldiskfs.<whatever>.quota_slave.enabled=<whatever>
*seems* to have done the trick!
[root at pgfs-mds3 ~]# lctl get_param osd-*.*.quota_slave.info
osd-ldiskfs.testfs-MDT0000.quota_slave.info=
target name: testfs-MDT0000
pool ID: 0
type: md
quota enabled: ug
conn to master: setup
space acct: ug
user uptodate: glb[1],slv[1],reint[0]
group uptodate: glb[1],slv[1],reint[0]
project uptodate: glb[0],slv[0],reint[0]
[root at pgfs-mds3 ~]# lctl get_param osd-ldiskfs.*.quota_slave.enabled
osd-ldiskfs.testfs-MDT0000.quota_slave.enabled=ug
[root at pgfs-mds3 ~]# lctl set_param
osd-ldiskfs.testfs-MDT0000.quota_slave.enabled=g
osd-ldiskfs.testfs-MDT0000.quota_slave.enabled=g
[root at pgfs-mds3 ~]# lctl get_param osd-*.*.quota_slave.info
osd-ldiskfs.testfs-MDT0000.quota_slave.info=
target name: testfs-MDT0000
pool ID: 0
type: md
quota enabled: g
conn to master: setup
space acct: ug
user uptodate: glb[1],slv[1],reint[0]
group uptodate: glb[1],slv[1],reint[0]
project uptodate: glb[0],slv[0],reint[0]
Another Q - if I use the 2.12+ default quota feature (ie sudo lfs
setquota -G -B 10m -I 1025 /testfs) and then read/write with an old
client will it enforce the quotas (even if I can't manipulate /
display them clearly) OK or is Something Bad (tm) going to happen
behind the scenes? - it _seemed_ to be behaving as expected from a
client side (2.7.mumble)
Many thanks
Andrew
More information about the lustre-discuss
mailing list