| Summary: | scsi_reserve: init script fails when device reports "unit attention" | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 5 | Reporter: | Ryan O'Hara <rohara> |
| Component: | cman | Assignee: | Ryan O'Hara <rohara> |
| Status: | CLOSED WONTFIX | QA Contact: | Cluster QE <mspqa-list> |
| Severity: | medium | Docs Contact: | |
| Priority: | low | ||
| Version: | 5.7 | CC: | cluster-maint, djansa, edamato, Sean.Stewart |
| Target Milestone: | rc | ||
| Target Release: | --- | ||
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Fixed In Version: | Doc Type: | Bug Fix | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | 640343 | Environment: | |
| Last Closed: | 2011-04-04 15:28:34 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Bug Depends On: | 640343 | ||
| Bug Blocks: | |||
|
Comment 1
Ryan O'Hara
2011-01-14 05:06:18 UTC
Changed BZ title since RHEL5 does not have unfencing. In RHEL5, the scsi_reserve init script is responsible for creating registrations and reservation. This init script should clear any "unit attention" state by issuing a TEST UNIT READY command to each device prior to creating registrations, checking status, etc. Closing this as WONTFIX. This bug was created is a clone of a bug that was fixed in RHEL6, but this is not really a problem for RHEL5. Whenever a node is fenced via fence_scsi, the device(s) will report "unit attention" (6) exactly once for any sg3_utils command issued after fencing completes. This is normal. Once "unit attention" has been reported for a device, subsequent sg3_utils commands (eg. sg_persist, sg_turs) will complete successfully. Normal, supported operation is that a node must be rebooted after being fenced via fence_scsi. After being rebooted, the node does not seem to reported "unit attention". Rebooting the fenced node seems to clear "unit attention". I was able to reproduce the problem using multipath devices and running "service scsi_reserve start" on the fenced node without rebooting. In this case, only the secondary paths would report "unit attention". It appears that the vgs command used to discover the devices will clear the unit attention state for the active paths. Rebooting the node appears to always clear "unit attention" state. Since rebooting the failed node is required, this bug is not something that needs to be fixed. |