Thanks Bob & Ovidiu.

From: Church, Robert [mailto:Robert.Church@windriver.com]
Sent: Wednesday, January 23, 2019 11:49 PM
To: Poncea, Ovidiu <Ovidiu.Poncea@windriver.com>; Liu, Changcheng <changcheng.liu@intel.com>; Kung, John <John.Kung@windriver.com>
Cc: starlingx-discuss@lists.starlingx.io
Subject: Re: [Starlingx-discuss] check_osds_down_up check range

If you run the command:

controller-0:~$ ceph osd tree --f json | python -c 'import json,sys;print json.load(sys.stdin)["nodes"][0]'

{u'children': [-2], u'type_id': 10, u'type': u'root', u'id': -1, u'name': u'storage-tier'}

The first entry is always the root entry, so we skip it.

ID WEIGHT  TYPE NAME                 UP/DOWN REWEIGHT PRIMARY-AFFINITY

-1 0.11620 root storage-tier

-2 0.11620     chassis group-0

-4 0.11620         host controller-0

0 0.11620             osd.0              up  1.00000          1.00000

-3       0         host controller-1

It’s a minor programmatic optimization.

Bob

From: Ovidiu Poncea <Ovidiu.Poncea@windriver.com>
Date: Wednesday, January 23, 2019 at 9:40 AM
To: "Liu, Changcheng" <changcheng.liu@intel.com>, "Kung, John" <John.Kung@windriver.com>
Cc: "starlingx-discuss@lists.starlingx.io" <starlingx-discuss@lists.starlingx.io>
Subject: Re: [Starlingx-discuss] check_osds_down_up check range

Hi Liu,

That code should work correctly as the first element in the list (index 0) is not of type 'host'.

Two notes here:

1. That code in cgcs-root/stx/stx-metal/inventory is not executed in a normal stx install as it is part of an sysinv in-progress refactoring. You should better look in cgcs-root/stx/stx-config/sysinv/sysinv/sysinv/sysinv/common/ceph.py

2. You shouldn't bother too much with that function (check_osds_down_up) as it was used at upgrades and we don't support upgrades in stx - it may get removed. Testing its output in small script (or even the python interpreter) should be enough at this stage.

Ovidiu

---

From: Liu, Changcheng [changcheng.liu@intel.com]
Sent: Wednesday, January 23, 2019 7:37 AM
To: Kung, John
Cc: starlingx-discuss@lists.starlingx.io
Subject: [Starlingx-discuss] check_osds_down_up check range

Hi John,

Could you help check whether there’s something wrong with below code?

   File: cgcs-root/stx/stx-metal/inventory/inventory/inventory/common/ceph.py

   96     def check_osds_down_up(self, hostname, upgrade):

   97         # check if osds from a storage are down/up

   98         response, body = self._ceph_api.osd_tree(body='json')

   99         osd_tree = body['output']['nodes']

  100         size = len(osd_tree)

  101         for i in range(1, size):

Is there something special reason to not range from 0? i.e. range(0, size)

B.R.

Changcheng