Bug 1471382 - IscsiDeploy can leave orphaned iscsi sessions
Summary: IscsiDeploy can leave orphaned iscsi sessions
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat OpenStack
Classification: Red Hat
Component: openstack-ironic
Version: 8.0 (Liberty)
Hardware: Unspecified
OS: Unspecified
medium
medium
Target Milestone: ---
: ---
Assignee: RHOS Maint
QA Contact: mlammon
URL:
Whiteboard:
Depends On:
Blocks: 1475433
TreeView+ depends on / blocked
 
Reported: 2017-07-15 12:31 UTC by Robin Cernin
Modified: 2021-02-01 02:38 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
: 1475433 (view as bug list)
Environment:
Last Closed: 2017-08-07 15:31:38 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Launchpad 1632649 0 None None None 2017-07-15 12:31:03 UTC
OpenStack gerrit 388092 0 None MERGED Cleanup hung iscsi session 2021-02-01 02:35:58 UTC

Description Robin Cernin 2017-07-15 12:31:03 UTC
Description of problem:

Discovered while debugging iSCSI deploy failures.

Deploy fails with:

"iSCSI connection was not seen by the file system after attempting to verify 3 times."

2017-07-14 17:47:18.271 1494 ERROR ironic.drivers.modules.agent_base_vendor   File "/usr/lib/python2.7/site-packages/ironic/drivers/modules/iscsi_deploy.py", line 364, in _fail_deploy
2017-07-14 17:47:18.271 1494 ERROR ironic.drivers.modules.agent_base_vendor     raise exception.InstanceDeployFailure(msg)
2017-07-14 17:47:18.271 1494 ERROR ironic.drivers.modules.agent_base_vendor InstanceDeployFailure: Deploy failed for instance f1c9feed-06f2-4f6b-bb40-683df39e90dd. Error: iSCSI connection was not seen by the file system after attempting to verify 3 times.

Running iscsiadm shows many iSCSI sessions

$ iscsiadm -m session
tcp: [10] xxxx:3260,1 iqn.2008-10.org.openstack:5255dc8a-0638-403b-83e7-5a7726d93873 (non-flash)
tcp: [100] xxxx:3260,1 iqn.2008-10.org.openstack:db9081a8-0aac-43d5-b792-03b964b20cb0 (non-flash)
tcp: [101] xxxx:3260,1 iqn.2008-10.org.openstack:bbe97782-ff4d-4f21-a533-1b36dfb973db (non-flash)
tcp: [102] xxxx:3260,1 iqn.2008-10.org.openstack:3e174b2e-855b-4e7d-aac7-df474664074b (non-flash)
...

These are presumably from attempts to deploy the nodes, but Ironic has not closed/cleaned up those sessions on deploy failure.

Version-Release number of selected component (if applicable):

openstack-ironic-api-4.2.5-3.el7ost.noarch                  Thu Jul  6 14:50:52 2017
openstack-ironic-common-4.2.5-3.el7ost.noarch               Thu Jul  6 14:42:24 2017
openstack-ironic-conductor-4.2.5-3.el7ost.noarch            Thu Jul  6 14:42:24 2017
openstack-ironic-inspector-2.2.6-1.el7ost.noarch            Thu Jul  6 14:50:47 2017
python-ironic-inspector-client-1.2.0-6.el7ost.noarch        Thu Jul  6 14:32:21 2017
python-ironicclient-0.8.1-1.el7ost.noarch                   Thu Jul  6 14:32:20 2017

Comment 2 Bob Fournier 2017-08-04 16:27:43 UTC
Hi Robin - it looks like there is a fix for this upstream as you've indicated in the external trackers. https://review.openstack.org/#/c/388092/.

Are you requesting a backport?

Comment 3 Bob Fournier 2017-08-07 15:31:38 UTC
Closing as fixed in current release.


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