RHBA-2018:2651 - Bug Fix Advisory

Topic

An update is now available for Red Hat Ceph Storage 2.5.

Description

Red Hat Ceph Storage is a scalable, open, software-defined storage platform
that combines the most stable version of the Ceph storage system with a
Ceph management platform, deployment utilities, and support services.

Bug Fix(es):

• Previously, the Ceph Object Gateway's Swift ACL processing logic was deficient in several respects. With this update, Swift ACL parsing has been enhanced, and as a result, RGW's Swift ACL conformance has been improved. In particular, public container ACLs can now be set. (BZ#1516099)
• Previously, the the result body for the Ceph Object Gateway's S3 "CompleteMultipartUpload" requests incorrectly omitted to return a "Location" member, but AWS S3 returns a value for "Location". With this update, the correct "Location" data is now returned from RGW when "CompleteMultipartUpload" completes successfully, and client workflows behave consistently for RGW and AWS S3. (BZ#1529727)
• "ceph-defaults" now contains all Ceph default variables, making "var_files" obsolete. Additionally, having "roles/ceph-defaults/defaults/main.yml" and "group_vars/all.yml" will create a collision and override necessary variables. (BZ#1555305)
• A bug in the "dir suggest" code caused inconsistent bucket and object stats when writing objects and listing objects concurrently on the bucket. With this update, the "dir suggest" code had been fixed to handle those cases, and now there are consistent bucket and objects stats.
• Ceph now supports the setting of Swift account and container quotas. BZ#15902400
• Placement groups (PGs) could become stuck in the "recovery_unfound" state when a copy of the data was available after an OSD was marked down and then up. To fix this issue, OSDs reset peering when they are marked down and then up, and PGs no longer get stuck in the "recovery_unfound" state. (BZ#1595283)
• The "radosgw-admin bucket check --fix" and bucket resharding could not fix mismatching stats resulting from inconsistent object stats. With this update, there is a new command "radosgw-admin bucket recalc-stats" that fixes the inconsistent object stats.

Note: This command has to be run when the bucket is offline. If there is multisite sync you need to stop all the zone radosgw daemons before running the command. (BZ#1609787)

• Previously, in Ceph Object Gateway multisite scenarios, an http request to another gateway would not complete and multisite sync would not make further progress. A timeout to the "libcurl" request has been added, and http requests that do not complete will time out and be retried, allowing multisite sync to continue.
• Previously, incorrect header-processing logic in the Ceph Object Gateway's S3 interface caused multi-part copy-part requests lacking the "x-amz-copy-source-range" header to be processed as whole-object copy requests. This caused requests of this type to fail when performed with AWSv4 request signing. With this update, the header processing logic has been corrected and S3 multi-part copy-part requests can be executed as expected by S3 clients. (BZ1614565)

Enhancement(s):

• A "trim delay" option has been added to the "radosgw-admin sync error trim" command in Ceph Object Gateway multisite. Previously, many OMAP keys could have been deleted by the full operation, leading to potential for impact on client workload. With the new option, trimming can be requested with low client workload impact.
• Increased detail has been added to the "radosgw-admin" sync status command, and a new "radosgw-admin bucket sync status" command has been added. The new commands provide more detailed reporting on the state of Ceph Object Gateway multi-site sync processes. Administrators can more easily confirm that multi-site sync is working correctly, and more easily debug sync issues.

Solution

Before applying this update, make sure all previously released errata
relevant to your system have been applied.

For details on how to apply this update, refer to:

https://access.redhat.com/articles/11258

Affected Products

• Red Hat Enterprise Linux Server 7 x86_64
• Red Hat Enterprise Linux Workstation 7 x86_64
• Red Hat Enterprise Linux Desktop 7 x86_64
• Red Hat Enterprise Linux for Scientific Computing 7 x86_64
• Red Hat Ceph Storage MON 2 x86_64
• Red Hat Ceph Storage OSD 2 x86_64

Fixes

• BZ - 1516099 - radosgw ACL are not correctly working with Swift CLI
• BZ - 1555305 - [CEE/SD][ceph-ansible][RHCS2] take-over-existing-cluster.yml fails with 'ceph_release_num.{{ ceph_release }} >= ceph_release_num.luminous'
• BZ - 1585700 - [CephFS] fusermount utility not available to unmount fuse mount
• BZ - 1595283 - [Backport] osd: eternal stuck PG in 'unfound_recovery' http://tracker.ceph.com/issues/24373
• BZ - 1599842 - segfault in cls_rgw
• BZ - 1601612 - [RFE] Add option --trim-delay-ms in radosgw-admin sync error trim command - to limit the frequency of osd ops
• BZ - 1613035 - [RFE] Filestore split log should show PG that is splitting
• BZ - 1614565 - jewel: rgw: copying part without http header "x-amz-copy-source-range" will be mistaken for copying object
• BZ - 1614926 - [RFE] RGW Multi-site radosgw-admin sync status and radosgw-admin bucket sync status commands addition of 'detail' flag
• BZ - 1619193 - enable multimds only on luminous

CVEs

(none)

References

(none)