CVE-2019-10222

A flaw was found in the Ceph RGW configuration with Beast as the front end handling client requests. An unauthenticated attacker could crash the Ceph RGW server by sending valid HTTP headers and terminating the connection, resulting in a remote denial of service for Ceph RGW clients.

These are all security issues fixed in the ceph-16.2.6.45+g8fda9838398-1.1 package on the GA media of openSUSE Tumbleweed.

This update for ceph fixes the following issues:

Security issue fixed:

• CVE-2019-10222: Fixed RGW crash via unauthenticated clients (bsc#1145093).

This update for ceph, ceph-iscsi and ses-manual_en fixes the following issues:

Security issues fixed:

• CVE-2019-10222: Fixed RGW crash caused by unauthenticated clients. (bsc#1145093)

Non-security issues-fixed:

• ceph-volume: prints errors to stdout with --format json (bsc#1132767)
• mgr/dashboard: Changing rgw-api-host does not get effective without disable/enable dashboard mgr module (bsc#1137503)
• mgr/dashboard: Silence Alertmanager alerts (bsc#1141174)
• mgr/dashboard: Fix e2e failures caused by webdriver version (bsc#1145759)
• librbd: always try to acquire exclusive lock when removing image (bsc#1149093)
• The no{up,down,in,out} related commands have been revamped (bsc#1151990)
• radosgw-admin gets two new subcommands for managing expire-stale objects. (bsc#1151991)
• Deploying a single new BlueStore OSD on a cluster upgraded to SES6 from SES5 breaks pool utilization stats reported by ceph df (bsc#1151992)
• Ceph cluster will no longer issue a health warning if CRUSH tunables are older than 'hammer' (bsc#1151993)
• Nautilus-based librbd clients can not open images on Jewel clusters (bsc#1151994)
• The RGW num_rados_handles has been removed in Ceph 14.2.3 (bsc#1151995)
• 'osd_deep_scrub_large_omap_object_key_threshold' has been lowered in Nautilus 14.2.3 (bsc#1152002)
• Support iSCSI target-level CHAP authentication (bsc#1145617)
• Validation and render of iSCSI controls based 'type' (bsc#1140491)
• Fix error editing iSCSI image advanced settings (bsc#1146656)
• Fix error during iSCSI target edit

Fixes in ses-manual_en:

• Added a new chapter with changelogs of Ceph releases. (bsc#1135584)
• Rewrote rolling updates and replaced running stage.0 with manual commands to prevent infinite loop. (bsc#1134444)
• Improved name of CaaSP to its fuller version. (bsc#1151439)
• Verify which OSD's are going to be removed before running stage.5. (bsc#1150406)
• Added two additional steps to recovering an OSD. (bsc#1147132)

Fixes in ceph-iscsi:

• Validate kernel LIO controls type and value (bsc#1140491)
• TPG lun_id persistence (bsc#1145618)
• Target level CHAP authentication (bsc#1145617)

ceph-iscsi was updated to the upstream 3.2 release:

• Always use host FQDN instead of shortname
• Validate min/max value for target controls and rbd:user/tcmu-runner image controls (bsc#1140491)

This update for ceph fixes the following issues:

• A previous update introduced a regression with the potential to cause RocksDB data corruption in Nautilus (bsc#1156282).

• Support for iSCSI target-level CHAP authentication was added (bsc#1145617).

• Implemented validation and rendering of iSCSI controls based 'type' (bsc#1140491).

• Fixed an error while editing iSCSI image advanced settings (bsc#1146656).

• Fixed a ceph-volume regression. SES customers were never exposed to this regression (bsc#1132767).

• Fixed a denial of service vulnerability where an unauthenticated client of Ceph Object Gateway could trigger a crash from an uncaught exception (bsc#1145093, CVE-2019-10222)

• Nautilus-based librbd clients could not open images on Jewel clusters (bsc#1151994).

• The RGW num_rados_handles has been removed (bsc#1151995).

• 'osd_deep_scrub_large_omap_object_key_threshold' has been lowered in Nautilus (bsc#1152002).

• The ceph dashboard now supports silencing Prometheus notifications (bsc#1141174).

• The no{up,down,in,out} related commands have been revamped (bsc#1151990).

• Radosgw-admin got two new subcommands for managing expire-stale objects (bsc#1151991)..

• Deploying a single new BlueStore OSD on a cluster upgraded to SES6 from SES5 used to break pool utilization stats reported by ceph df (bsc#1151992).

• Ceph clusters will issue a health warning if CRUSH tunables are older than 'hammer' (bsc#1151993).

• Ceph-volume prints errors to stdout with --format json (bsc#1132767).

• Changing rgw-api-host in the dashboard does not get effective without disable/enable dashboard mgr module (bsc#1137503).

• Silenced Alertmanager alerts in the dashboard (bsc#1141174).

• Fixed e2e failures in the dashboard caused by webdriver version (bsc#1145759)

• librbd always tries to acquire exclusive lock when removing image an (bsc#1149093).

Fixes in ses-manual_en:

• Added a new chapter with changelogs of Ceph releases. (bsc#1135584)
• Rewrote rolling updates and replaced running stage.0 with manual commands to prevent infinite loop. (bsc#1134444)
• Improved name of CaaSP to its fuller version. (bsc#1151439)
• Verify which OSD's are going to be removed before running stage.5. (bsc#1150406)
• Added two additional steps to recovering an OSD. (bsc#1147132)

Fixes in ceph-iscsi:

• Validate kernel LIO controls type and value (bsc#1140491)
• TPG lun_id persistence (bsc#1145618)
• Target level CHAP authentication (bsc#1145617)

ceph-iscsi was updated to the upstream 3.2 release:

• Always use host FQDN instead of shortname
• Validate min/max value for target controls and rbd:user/tcmu-runner image controls (bsc#1140491)

Abhishek Lekshmanan discovered that the RADOS gateway implementation in Ceph did not handle client disconnects properly in some situations. A remote attacker could use this to cause a denial of service.