Hi Juan,

I confirm this was working with 4.0.6 (but the same SDK package). The script is relatively new but I'm 100% sure it has been in execution for several weeks, in contrast to current version where after 5-6 iterations the exception starts showing up.

I've voted for the bug and added myself as CC.

Thank you.

Regards,

Nicolás

El 2017-03-22 12:58, Juan Hernández escribió:

On 03/22/2017 01:50 PM, Juan Hernández wrote:

On 03/22/2017 01:10 PM, nicolas@devels.es wrote:

Hi,

We've recently upgraded from oVirt 4.0.x to 4.1.0 and it seems that the
behavior of a script we wrote in Python-SDK has changed slightly. We
have a script that needs to be executed forever (daemon mode). This
daemon creates an ovirtsdk4.Connection object and uses the same
Connection object all the time.

conn = sdk.Connection(
  url=URI,
  username=USERNAME,
  password=PASSWORD,
  ca_file=CAFILE
)

Between iterations we have ~5 minutes delays which we accomplish with a
time.sleep(...) call.

After some of these iterations (5 or 6), when trying to perform an
operation on the Connection object (in this case, listing all SDs):

    try:
        sys_serv = conn.system_service()
        sd_serv = sys_serv.storage_domains_service()
        storages = sd_serv.list(search=sd_search_query)
    except Error, e:
        log('ERR: Error getting storage domains: %s' % (e))

We start getting exceptions (i.e., the sd_serv.list() call throwing an
Error exception) with this message:

    ERR: Error getting storage domains: HTTP response code is 401.

I believe this has something to do with the Connection object expiring
(even if it's not idle more than 5 minutes at any time). Is there a way
to "refresh" the Connection object so it doesn't auto-logout (assuming
this is the actual problem)?

As I said, this started happening as of oVirt 4.1.0.
ovirt-engine-sdk-python version is 4.1.1.

Any ideas?

This sound like this bug, opened yesterday:

  SSO token used for the API expires when running only queries
  https://bugzilla.redhat.com/1434605

But I thought the same happened in 4.0. Can you confirm that you don't
see this problem in 4.0?

Open possible workaround is to force a refresh of the backend session
sending an external event, like in this example:


https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/vm_backup.py#L112-L131

I also think that we need to modify the SDKs so that they detect expired
SSO tokens and renew them automatically. I will open another bug for that.

Here is the bug for adding automatic SSO token renew to the SDK, in case
you want to follow/vote it:

  Implement automatic SSO token renew
  https://bugzilla.redhat.com/1434830

_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users