Open whowutwut opened 4 years ago
It seems like there's no scale_storage
when we get into https://github.com/IBM/ibm-spectrum-scale-install-infra/blob/dev/roles/core/node/templates/StanzaFile.j2#L1
I cleared out this file and just added...
debug: {{ current_fs }}
debug: {{ scale_storage }}
And the following is printed:
debug: gpfs01
debug - scale_storage: []
Just for my own sanity, I checked out 6bbb0be5e0ef162711321566c9167c92b5aa0543 which is really early on in master... ran the same files through it... and NSDs get created..
[root@autogen-hostvars-rhels77-x-master ~]# cat /var/tmp/StanzaFile.new.gpfs01
%nsd:
device=/dev/vdb
nsd=nsd_autogen-hostvars-rhels77-x-worker1_vdb
servers=autogen-hostvars-rhels77-x-worker1
usage=dataAndMetadata
failureGroup=2
pool=system
%nsd:
device=/dev/vdc
nsd=autogen_hostvars_rhels77_x_worker1_nsd_vdc
servers=autogen-hostvars-rhels77-x-worker1
usage=dataAndMetadata
failureGroup=2
pool=system
%nsd:
device=/dev/vdb
nsd=nsd_autogen-hostvars-rhels77-x-worker2_vdb
servers=autogen-hostvars-rhels77-x-worker2
usage=dataAndMetadata
failureGroup=2
pool=system
%nsd:
device=/dev/vdc
nsd=autogen_hostvars_rhels77_x_worker2_nsd_vdc
servers=autogen-hostvars-rhels77-x-worker2
usage=dataAndMetadata
failureGroup=2
pool=system
Edit: I have to find a better commit to check out, this one I picked creates the NSDs in the Stanza, (which at least proves that my host_vars files does work)... but I have issues with passing in -
to the nsd names, which was fixed later on in the code.
@whowutwut i think it might be servers and nsd name in the group_vars/all that you have added in double quotes creating issue, remove double quotes then it should work. Yours definition ,
disks:
- device: /dev/vdb
servers: "autogen-hostvars-rhels77-x-worker1"
- device: /dev/vdc
nsd: "autogen_hostvars_rhels77_x_worker1_nsd_vdc"
servers: "autogen-hostvars-rhels77-x-worker1
@whowutwut I realized one more thing, You have defined this parameter in the host_vars, now it will not work with host_vars , user always needs to defined in the group_vars.
README statement:
Important: scale_storage
must be define using group_vars
inventory files. Do not define disk parameters using host_vars inventory files or inline
We already had a multiple discussion in the slack channel about group_vars and host_vars.
We already had a multiple discussion in the slack channel about group_vars and host_vars.
Sure, understand. But I hit this again, and did not realize what was going on. So other reasons for opening issues in this public tracker is that it helps the community. If anyone out there runs into similar things, these issues will appear in google searches once the crawlers do their thing.
host_vars
and switch to defining scale_storage under group_vars
, which include the NSD definitions.
Need some help... I'm backed to the current dev branch (also tried master)... . using host_vars, where I have disks on workers, seems to be skipped in creating the NSD.
Is there something wrong with my config? I do think this was working in the past... How to best debug this?
Here's my host_vars file:
I tried it both with the content of this file and removing those lines indicated, no changes, still skipped NSDs..
But then I see the tasks being skipped....
and then after the playbook is completed
On another cluser, using group_vars with this config works, no problem