After the timeout, an attempt was made to unshelve the instance via the /unshelve comment. However, because the server status remained as ACTIVE due to the earlier timeout, the unshelve command incorrectly assumed success. This occurred because the workflow does not currently verify the "OS-EXT-STS:task_state" server property, which shows intermediate states during status transitions.
For more details on task states and server statuses, please refer to the following documentation:
Verify pending transitions before assuming a command's success, ensuring that the server status is correctly updated before proceeding with subsequent commands.
Consider increasing the polling timeout for control commands.
jcfr
commented
1 month ago
When the polling associated with control commands like
shelvetimes out, subsequent commands may improperly report success.For instance, the
shelvecommand executed on the server timed out before the server status could update fromACTIVEtoSHELVED_OFFLOADED.This issue can be observed in the following links:
After the timeout, an attempt was made to
unshelvethe instance via the/unshelvecomment. However, because the server status remained asACTIVEdue to the earlier timeout, theunshelvecommand incorrectly assumed success. This occurred because the workflow does not currently verify the "OS-EXT-STS:task_state" server property, which shows intermediate states during status transitions.For more details on task states and server statuses, please refer to the following documentation:
Proposed Solution: