Bug 1646872
| Summary: | [OSP] SBD cannot be used with bundles [rhel-7.6.z] | ||
|---|---|---|---|
| Product: | Red Hat Enterprise Linux 7 | Reporter: | Oneata Mircea Teodor <toneata> |
| Component: | pacemaker | Assignee: | Ken Gaillot <kgaillot> |
| Status: | CLOSED ERRATA | QA Contact: | Marian Krcmarik <mkrcmari> |
| Severity: | urgent | Docs Contact: | |
| Priority: | urgent | ||
| Version: | 7.6 | CC: | abeekhof, aherr, cfeist, cluster-maint, ctowsley, kgaillot, kwenning, michele, mjuricek, mkrcmari, sbradley |
| Target Milestone: | rc | Keywords: | Triaged, ZStream |
| Target Release: | 7.6 | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
| Whiteboard: | |||
| Fixed In Version: | pacemaker-1.1.19-8.el7_6.1 | Doc Type: | Bug Fix |
| Doc Text: |
Cause: When SBD is configured on the cluster nodes, Pacemaker Remote nodes (including guest nodes and bundle nodes) will compare the local SBD configuration and abort if not compatible.
Consequence: Guest nodes and bundle nodes unnecessarily fail when SBD is used on the cluster nodes, since they use resource recovery rather than standard fencing mechanisms.
Fix: Pacemaker Remote skips the SBD compatibility check when run on a guest node or bundle node.
Result: Guest nodes and bundle nodes may be used in a cluster with SBD, without configuring SBD on the guest nodes or bundle nodes themselves.
|
Story Points: | --- |
| Clone Of: | 1638593 | Environment: | |
| Last Closed: | 2018-11-27 01:21:58 UTC | Type: | --- |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
| Bug Depends On: | 1638593 | ||
| Bug Blocks: | |||
|
Description
Oneata Mircea Teodor
2018-11-06 08:08:48 UTC
Fixed in upstream 1.1 branch by commit 4dae674 QA: To reproduce, configure a cluster with a guest node and/or bundle, where all the cluster nodes use sbd for fencing. Configure a resource to run on the guest node and/or in the bundle that will require contacting the CIB or controller, such as galera, redis, or rabbitmq (or a modified Dummy resource that calls something like cibadmin -Q). Before the fix, the guest node and/or bundle instance will get restarted when the resource starts on it; after the fix, it will proceed normally. I am going to switch to Verified even though I do not have clear knowledge of what to verify here and how to reproduce, If I configure watchdog with softdog and sbd and I set stonith-watchdog-timeout property then All the bundle resource are stopped on the setup with unfixed pacemaker, once I use fixed version of pacemaker it starts to work. I have no see any unneccessary restarts of bundles in any case though Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2018:3667 |