Correct setup of the layout

Site Administrators
1

Dear All,

Because of the relatively old disks we have, I would like to set up our instance to have a fully mirrored layout, in order to increase safety. These are the commands I performed:

[root@eos-mgm1 ~]# eos space config default space.policy.layout=replica
[root@eos-mgm1 ~]# eos space config default space.policy.nstripes=2
[root@eos-mgm1 ~]# eos attr set default=replica /eos      
[root@eos-mgm1 ~]# eos attr ls eos
sys.forced.blocksize="4k"
sys.forced.checksum="adler"
sys.forced.layout="replica"
sys.forced.nstripes="2"
sys.forced.space="default"

My question is that is this enough? When I list the space info, I still see the full storage capacity instead of only the half of it (sum(capacity)=567.04 TB and capacity(rw)=566.92 TB, along with nom.capacity=0 B).

Currently the system consists of 8 FSTs, with 24*3 TB disks each.

Thanks,
Gabor

2

Since the layout is set on each directory, the ‘space’ command always shows the physical space.
Only quota shows logical space according to the layout defined on the quota node.

What you did is sufficient, but if you have already directories under /eos/, you should set it also there.

You can use the recursive option to do that, but avoid to set this settings into the /eos/…/proc/ directory.

Cheers Andreas.

3

Hi Andreas,

Thanks for the answer. Regarding the quota: currently it is set to off. Is it enought to simply turn it on?

Cheers,
Gabor

4

You do:

# enable the quota
eos space quota default on 
# create a project quota (done by using group 99 to 280TB logical space and 100M files)
eos quota set -g 99 -v 280T -i 100M

# check the quota
eos quota ls
5

Thanks for the clarification again. However, there is still one thing: I apllied this and the ls says that the 280T is the physical space, not the logical one:

[root@eos-mgm1 ~]# eos quota ls
# pre-configuring default route to /eos/user/r/root/
# -use $EOSHOME variable to override

┏━> Quota Node: /eos/alice/grid/
┌──────────┬──────────┬──────────┬──────────┬──────────┬──────────┬──────────┬──────────┬──────────┬──────────┐
│group     │used bytes│logi bytes│used files│aval bytes│aval logib│aval files│ filled[%]│vol-status│ino-status│
└──────────┴──────────┴──────────┴──────────┴──────────┴──────────┴──────────┴──────────┴──────────┴──────────┘
 project           0 B        0 B          0  280.00 TB  140.00 TB   100.00 M     0.00 %         ok         ok 
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛

So, can I just set the quota with a higher value, or do I have to first remove this ‘old’ quota? This is similar to the situation when e.g. additional, new FST nodes are added.

Thanks,
Gabor