Bug 1130038 - prepareImage api call fails with [Errno 2] No such file or directory
Summary: prepareImage api call fails with [Errno 2] No such file or directory
Keywords:
Status: CLOSED NOTABUG
Alias: None
Product: oVirt
Classification: Retired
Component: vdsm
Version: 3.5
Hardware: Unspecified
OS: Unspecified
unspecified
urgent
Target Milestone: ---
: 3.5.0
Assignee: Federico Simoncelli
QA Contact: Gil Klein
URL:
Whiteboard: sla
Depends On:
Blocks: 1067162 1073943
TreeView+ depends on / blocked
 
Reported: 2014-08-14 08:25 UTC by Jiri Moskovcak
Modified: 2016-02-10 19:42 UTC (History)
14 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of: 1129261
Environment:
Last Closed: 2014-08-25 15:30:52 UTC
oVirt Team: SLA
Embargoed:


Attachments (Terms of Use)
logs and test (171.05 KB, application/zip)
2014-08-21 07:45 UTC, Jiri Moskovcak
no flags Details

Description Jiri Moskovcak 2014-08-14 08:25:53 UTC
+++ This bug was initially created as a clone of Bug #1129261 +++

Description of problem:
when hosted engine broker tries to connect it's storage using vdsm it fail with: [Errno 2] No such file or directory

Version-Release number of selected component (if applicable):
vdsm-4.16.1-6.gita4a4614.el6.x86_64

How reproducible:
100%

Steps to Reproduce:
1. install hosted-engine
2. reboot the machine
3. check if the hosted engine storage is up'n'running
4. /rhev/data-center/mnt/blockSD/3498465f-eb04-49f3-81d2-f7713b392e5a/ha_agent/hosted-engine.metadata points to volume which is not activated

Actual results:
ha-agent doesn't run

Expected results:
storage connected, ha-agent running

Additional info:
vdsm log attached

--- Additional comment from Jiri Moskovcak on 2014-08-12 08:28:11 EDT ---

after additional testing, it seems like it actually activates the lvm device, but fails to create some symlink and returns non-zero error code which makes the he-broker think that it fails even though the storage is actually available.

--- Additional comment from Dan Kenigsberg on 2014-08-12 10:21:50 EDT ---

Thread-1205::ERROR::2014-08-12 14:13:03,605::blockSD::1054::Storage.StorageDomain::(linkBCImage) Failed to create path to image directory: /rhev/data-center/84cfc3d9-03cf-4ec5-9f29-a18ba1f6f02f/3498465f-eb04-49f3-81d2-f7713b392e5a/images/04f6b652-5e14-40d8-973c-ca384ffc0c50
Thread-1205::ERROR::2014-08-12 14:13:03,605::task::866::Storage.TaskManager.Task::(_setError) Task=`0ffda0cc-8afc-46bb-8b80-49e0a4291a68`::Unexpected error
Traceback (most recent call last):
  File "/usr/share/vdsm/storage/task.py", line 873, in _run
    return fn(*args, **kargs)
  File "/usr/share/vdsm/logUtils.py", line 45, in wrapper
    res = f(*args, **kwargs)
  File "/usr/share/vdsm/storage/hsm.py", line 3213, in prepareImage
    runImgPath = dom.linkBCImage(imgPath, imgUUID)
  File "/usr/share/vdsm/storage/blockSD.py", line 1047, in linkBCImage
    os.symlink(imgPath, dst)
OSError: [Errno 2] No such file or directory

--- Additional comment from Federico Simoncelli on 2014-08-13 11:52:06 EDT ---

The request fails because you're not connected to a pool. To prepare an image when you're not connected to a pool (general HE use case) you should specify a blank pool uuid (00000000-0000-0000-0000-000000000000). E.g.:

prepareImage(sdUUID='3498465f-eb04-49f3-81d2-f7713b392e5a', spUUID='00000000-0000-0000-0000-000000000000', imgUUID='e3449f52-b227-4666-a4d5-9fdc56f000ce', leafUUID='04f6b652-5e14-40d8-973c-ca384ffc0c50')

--- Additional comment from Jiri Moskovcak on 2014-08-14 03:33:25 EDT ---

(In reply to Federico Simoncelli from comment #3)
> The request fails because you're not connected to a pool. To prepare an
> image when you're not connected to a pool (general HE use case) you should
> specify a blank pool uuid (00000000-0000-0000-0000-000000000000). E.g.:
> 
> prepareImage(sdUUID='3498465f-eb04-49f3-81d2-f7713b392e5a',
> spUUID='00000000-0000-0000-0000-000000000000',
> imgUUID='e3449f52-b227-4666-a4d5-9fdc56f000ce',
> leafUUID='04f6b652-5e14-40d8-973c-ca384ffc0c50')

- thanks for the hint, going to try that, btw, I'm pretty sure it used to work back in May when we start using vdsm api for storage creation.

--- Additional comment from Jiri Moskovcak on 2014-08-14 03:47:59 EDT ---

btw, does the same apply for getVolumePath ?

--- Additional comment from Jiri Moskovcak on 2014-08-14 04:06:09 EDT ---

This is what I get if I use blank pool:

Thread-199::ERROR::2014-08-14 10:04:35,846::task::866::Storage.TaskManager.Task::(_setError) Task=`95689457-eec0-405c-8d0c-3f81e2556121`::Unexpected error
Traceback (most recent call last):
  File "/usr/share/vdsm/storage/task.py", line 873, in _run
    return fn(*args, **kargs)
  File "/usr/share/vdsm/logUtils.py", line 45, in wrapper
    res = f(*args, **kwargs)
  File "/usr/share/vdsm/storage/hsm.py", line 3237, in prepareImage
    return {'path': leafPath, 'info': leafInfo,
UnboundLocalError: local variable 'leafInfo' referenced before assignment

Comment 1 Jiri Moskovcak 2014-08-21 07:45:16 UTC
Created attachment 929080 [details]
logs and test

Comment 2 Federico Simoncelli 2014-08-25 09:54:56 UTC
Please report the output of:

 # lvs -o +tags

Comment 3 Federico Simoncelli 2014-08-25 10:15:45 UTC
(In reply to Federico Simoncelli from comment #2)
> Please report the output of:
> 
>  # lvs -o +tags

No need for that I just saw that you inverted imgUUID and volUUID:

conn = vdscli.connect()
print "connecting metadata"
response = conn.prepareImage(
       sp_uuid,
       sd_uuid,
       metadata_volume_UUID,
       metadata_image_UUID
)

print response
print "connecting lockspace"
reposnse = conn.prepareImage(
       sp_uuid,
       sd_uuid,
       lockspace_volume_UUID,
       lockspace_image_UUID
)

Should be:

response = conn.prepareImage(
       sp_uuid,
       sd_uuid,
       metadata_image_UUID,
       metadata_volume_UUID
)

reposnse = conn.prepareImage(
       sp_uuid,
       sd_uuid,
       lockspace_image_UUID,
       lockspace_volume_UUID
)

Comment 4 Federico Simoncelli 2014-08-25 15:30:52 UTC
Closing as the error was caused from the API misuse.

To improve the error message we'll use bug 1133617.


Note You need to log in before you can comment on or make changes to this bug.