Glance task creates failed when setting work_dir local and qemu-img version is 1.5.3

[Yafei Yu]

The openstack version is mitaka.

# rpm -qa |grep qemu-img
qemu-img-1.5.3-105.el7_2.4.x86_64

The glance-api.conf setting is:
[task]
work_dir = /home/work/
[taskflow_executor]
conversion_format = raw

Then run the cli:
glance task-create --type import --input '{"import_from":"http://10.43.177.17/cirros-0.3.2-x86_64-disk.img","import_from_format": "","image_properties":{"disk_format":"qcow2","container_format":"bare","name":"test1"}}'

The log is :
2016-06-14 04:08:29.032 DEBUG oslo_concurrency.processutils [-] CMD "qemu-img info --output=json file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d" returned: 1 in 0.025s from (pid=5460) execute /usr/lib/python2.7/site-packages/oslo_concurrency/processutils.py:374
2016-06-14 04:08:29.033 DEBUG oslo_concurrency.processutils [-] None
command: u'qemu-img info --output=json file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d'
exit code: 1
stdout: u''
stderr: u"qemu-img: Could not open 'file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d': Could not open 'file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d': No such file or directory\n" from (pid=5460) execute /usr/lib/python2.7/site-packages/oslo_concurrency/processutils.py:413
2016-06-14 04:08:29.034 DEBUG oslo_concurrency.processutils [-] u'qemu-img info --output=json file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d' failed. Not Retrying. from (pid=5460) execute /usr/lib/python2.7/site-packages/oslo_concurrency/processutils.py:422
Command: qemu-img info --output=json file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d
Exit code: 1
Stdout: u''
Stderr: u"qemu-img: Could not open 'file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d': Could not open 'file:///home/work/90ff2129-0079-487e-a7ec-79ef23bd1c0d': No such file or directory\n"
2016-06-14 04:08:29.072 WARNING glance.async.taskflow_executor [-] Task 'import-ImportToFS-42684807-86db-4ff5-a4a9-abf3b1998b63' (5ff9cf63-f257-48d2-9cc9-cfeffd905854) transitioned into state 'FAILURE' from state 'RUNNING'
4 predecessors (most recent first):
  Flow 'import'
  |__Atom 'import-CreateImage-42684807-86db-4ff5-a4a9-abf3b1998b63' {'intention': 'EXECUTE', 'state': 'SUCCESS', 'requires': {}, 'provides': '90ff2129-0079-487e-a7ec-79ef23bd1c0d'}
     |__Atom 'import_retry' {'intention': 'EXECUTE', 'state': 'SUCCESS', 'requires': {}, 'provides': [(None, {})]}
        |__Flow 'import'
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor Traceback (most recent call last):
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor File "/usr/lib/python2.7/site-packages/taskflow/engines/action_engine/executor.py", line 82, in _execute_task
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor result = task.execute(**arguments)
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor File "/opt/stack/glance/glance/async/flows/base_import.py", line 175, in execute
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor metadata = json.loads(stdout)
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor File "/usr/lib64/python2.7/json/__init__.py", line 338, in loads
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor return _default_decoder.decode(s)
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor File "/usr/lib64/python2.7/json/decoder.py", line 365, in decode
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor obj, end = self.raw_decode(s, idx=_w(s, 0).end())
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor File "/usr/lib64/python2.7/json/decoder.py", line 383, in raw_decode
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor raise ValueError("No JSON object could be decoded")
2016-06-14 04:08:29.072 TRACE glance.async.taskflow_executor ValueError: No JSON object could be decoded

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/329394

[Nikhil Komawar (nikhil-komawar)]

I think the bug seems to be in oslo_concurrency.processutils because glance is merely using:

---

glance/async/flows/base_import.py: stdout, stderr = putils.trycmd('qemu-img', 'info',
glance/async/flows/convert.py: # format already. Probably using `qemu-img` just like the
glance/async/flows/convert.py: stdout, stderr = putils.trycmd('qemu-img', 'convert', '-O',
glance/async/flows/introspect.py: stdout, stderr = putils.trycmd('qemu-img', 'info',

---

So, the file:/ should not appear in the task command because glance does not set those. Can you reconfirm?

[Yafei Yu (yu-yafei)]

Thank you, Nikhil.
I make a test in two host, and the result looks because different versions of qemu-img.
if path is "file:///work_dir", qemu-img command failed, but if "/work_dir" qemu-img works well when qemu-img-1.5.3-105.el7_2.4.x86_64.
So I make a change in glance:"file:///work_dir"--->"/work_dir"

[root@stack ~]# rpm -qa |grep release
isorelax-0-0.15.release20050331.el7.noarch
centos-release-7-0.1406.el7.centos.2.3.x86_64
rdo-release-mitaka-2.noarch
epel-release-7-6.noarch
[root@stack ~]# rpm -qa |grep qemu-img
qemu-img-1.5.3-105.el7_2.4.x86_64
[root@stack ~]# qemu-img info --output=json file:///home/cirros-0.3.1-x86_64-disk.img
qemu-img: Could not open 'file:///home/cirros-0.3.1-x86_64-disk.img': Could not open 'file:///home/cirros-0.3.1-x86_64-disk.img': No such file or directory
[root@stack ~]# qemu-img info --output=json /home/cirros-0.3.1-x86_64-disk.img
{
    "virtual-size": 41126400,
    "filename": "/home/cirros-0.3.1-x86_64-disk.img",
    "cluster-size": 65536,
    "format": "qcow2",
    "actual-size": 13148160,
    "format-specific": {
        "type": "qcow2",
        "data": {
            "compat": "0.10"
        }
    },
    "dirty-flag": false
}

[root@2C5_10_DELL05 ~]# rpm -qa |grep release
centos-release-7-0.1406.el7.centos.2.4.x86_64
tfg-release-1.0-472.6.1.15.noarch
[root@2C5_10_DELL05 ~]# rpm -qa |grep qemu-img
qemu-img-2.5.0-5.6.1.1.el7.centos.x86_64
[root@2C5_10_DELL05 ~]# qemu-img info --output=json file:///home/cirros-0.3.1-x86_64-disk.img
{
    "virtual-size": 41126400,
    "filename": "///home/cirros-0.3.1-x86_64-disk.img",
    "cluster-size": 65536,
    "format": "qcow2",
    "actual-size": 13148160,
    "format-specific": {
        "type": "qcow2",
        "data": {
            "compat": "0.10",
            "refcount-bits": 16
        }
    },
    "dirty-flag": false
}
[root@2C5_10_DELL05 ~]# qemu-img info --output=json /home/cirros-0.3.1-x86_64-disk.img
{
    "virtual-size": 41126400,
    "filename": "/home/cirros-0.3.1-x86_64-disk.img",
    "cluster-size": 65536,
    "format": "qcow2",
    "actual-size": 13148160,
    "format-specific": {
        "type": "qcow2",
        "data": {
            "compat": "0.10",
            "refcount-bits": 16
        }
    },
    "dirty-flag": false
}

[OpenStack Infra (hudson-openstack)]

Change abandoned by yuyafei (<email address hidden>) on branch: master
Review: https://review.openstack.org/329394

[Brian Rosmaita (brian-rosmaita)]

Looks like this was fixed by configuration, closing.