Back to bug 2149468

Who When What Removed Added
RHEL Program Management 2022-11-30 00:54:38 UTC Target Release --- 17.1
melanie witt 2022-11-30 00:57:38 UTC Keywords Patch
Link ID OpenStack gerrit 866085
Status ASSIGNED ON_DEV
Red Hat One Jira (issues.redhat.com) 2022-11-30 01:15:33 UTC Link ID Red Hat Issue Tracker OSP-20556
melanie witt 2022-11-30 01:20:03 UTC Link ID OpenStack gerrit 866084
melanie witt 2022-11-30 01:20:54 UTC Blocks 2149471
melanie witt 2022-12-07 02:42:36 UTC Blocks 2151410
errata-xmlrpc 2023-01-11 13:37:27 UTC Fixed In Version openstack-nova-23.2.3-1.20221209191244.bbf626c.el9ost
Status ON_DEV MODIFIED
CC smooney
Status MODIFIED ON_QA
Artom Lifshitz 2023-03-21 20:00:34 UTC QA Contact nova-maint osp-dfg-compute
Andy Stillman 2023-05-31 21:04:30 UTC Flags needinfo?(mwitt)
melanie witt 2023-06-07 21:04:50 UTC Doc Type If docs needed, set a value Bug Fix
Doc Text Cause: Temporary error response from cinder service volume detach API such as 504 Gateway Timeout.

Consequence: Nova considers it an error and fails the volume detach even though it succeeded (but timed out) on the cinder side, leaving a stale block device mapping record in the nova database.

Fix: Make nova retry the volume detach call to the cinder API if it receives an HTTP error that is likely to be temporary. Upon retry, if the volume attachment is no longer found, nova will consider it already detached (success).

Result: Nova will handle temporary error conditions and retry to succeed with volume detach in those cases.
melanie witt 2023-06-07 21:05:53 UTC Flags needinfo?(mwitt)
Joanne O'Flynn 2023-06-08 08:29:12 UTC Flags needinfo?(mwitt)
CC joflynn
Doc Text Cause: Temporary error response from cinder service volume detach API such as 504 Gateway Timeout.

Consequence: Nova considers it an error and fails the volume detach even though it succeeded (but timed out) on the cinder side, leaving a stale block device mapping record in the nova database.

Fix: Make nova retry the volume detach call to the cinder API if it receives an HTTP error that is likely to be temporary. Upon retry, if the volume attachment is no longer found, nova will consider it already detached (success).

Result: Nova will handle temporary error conditions and retry to succeed with volume detach in those cases.
Before this update, a temporary error response from the Block Storage service (cinder) volume detach API, such as 504 Gateway Timeout, was considered by the Compute service (nova) as an error. The Compute service failed the volume detach operation even though it succeeded but timed out on the Block Storage service side, leaving a stale block device mapping record in the Compute service database. With this update, the Compute service retries the volume detach call to the Block Storage service API if it receives an HTTP error that is likely to be temporary. Upon retry, if the volume attachment is no longer found, the Compute service considers the volume already detached.
melanie witt 2023-06-08 16:19:39 UTC Flags needinfo?(mwitt)
Jenny-Anne Lynch 2023-06-14 14:31:37 UTC Doc Text Before this update, a temporary error response from the Block Storage service (cinder) volume detach API, such as 504 Gateway Timeout, was considered by the Compute service (nova) as an error. The Compute service failed the volume detach operation even though it succeeded but timed out on the Block Storage service side, leaving a stale block device mapping record in the Compute service database. With this update, the Compute service retries the volume detach call to the Block Storage service API if it receives an HTTP error that is likely to be temporary. Upon retry, if the volume attachment is no longer found, the Compute service considers the volume already detached. Before this update, the Compute service (nova) processed a temporary error message from the Block Storage service (cinder) volume detach API, such as '504 Gateway Timeout', as an error. The Compute service failed the volume detach operation even though it succeeded but timed out on the Block Storage service side, leaving a stale block device mapping record in the Compute service database. With this update, the Compute service retries the volume detach call to the Block Storage service API if it receives an HTTP error that is likely to be temporary. Upon retry, if the volume attachment is no longer found, the Compute service processes the volume as already detached.
CC jelynch
Jenny-Anne Lynch 2023-07-19 12:19:48 UTC CC jelynch
errata-xmlrpc 2023-08-16 00:04:07 UTC Status ON_QA RELEASE_PENDING
errata-xmlrpc 2023-08-16 01:12:55 UTC Status RELEASE_PENDING CLOSED
Resolution --- ERRATA
Last Closed 2023-08-16 01:12:55 UTC
errata-xmlrpc 2023-08-16 01:13:41 UTC Link ID Red Hat Product Errata RHEA-2023:4577

Back to bug 2149468