Release Notes #

ceph-mgr

python-influxdb

The command syntax could be intimidating.

Error messages from Salt can be difficult to understand.

Some DeepSea operations triggered by commands require a longer time to complete. The lack of feedback made it difficult to determine whether DeepSea is still working or has aborted the operation.

Cloning an RBD snapshot creates a new RBD image by entering a new name for the new clone.

Optionally, you can select different feature set for the new RBD image.

You can also copy RBD images, which can be helpful when several RBD images with the exact same size and feature(s) are needed (but as in he case of cloning, features can be also added).

To avoid accidentally deleting a snapshot of an RBD image, you can mark it as protected.

RHEL 6 and 7

Ubuntu 16.04 and 18.04

SUSE Enterprise Storage Administration Guide, Chapter “Stored Data Management”, Section “Devices”, Section “Migrating from a Legacy SSD Rule to Device Classes” at https://www.suse.com/documentation/suse-enterprise-storage-5/singlehtml/book_storage_admin/book_storage_admin.html#device_classes.reclassify

The upstream documentation at http://docs.ceph.com/docs/luminous/rados/operations/crush-map-edits/#migrating-from-a-legacy-ssd-rule-to-device-classes

Ceph now defaults to the AsyncMessenger (ms type async) instead of the legacy SimpleMessenger. The most noticeable difference is that Ceph now uses a fixed-sized thread pool for network connections instead of two threads per socket with SimpleMessenger.

Some OSD failures are now detected almost immediately, whereas previously the heartbeat timeout (which defaults to 20 seconds) had to expire. This prevents I/O from blocking for an extended period for failures where the host remains up but the ceph-osd process is no longer running.

The size of encoded OSDMaps has been reduced.

The OSDs now quiesce scrubbing when recovery or rebalancing is in progress.

For Luminous clients, there is a new upmap exception mechanism that allows individual Placement Groups (PGs) to be moved around to achieve a perfect distribution.

Each OSD now adjusts its default configuration based on whether the backing device is an hard disk or SSD. Manual tuning is generally not required.

There is now a back-off mechanism that prevents OSDs from being overloaded by requests to objects or PGs that currently cannot process IO.

There is a simplified OSD replacement process that is more robust.

You can query the supported features and (apparent) releases of all connected daemons and clients with Ceph features.

You can configure the oldest Ceph client version you want to allow to connect to the cluster via ceph osd set-require-min-compat-client. Ceph will then prevent you from enabling features that will break compatibility with those clients.

Several sleep settings, including osd_recovery_sleep, osd_snap_trim_sleep, and osd_scrub_sleep have been reimplemented to work more efficiently.

Pools are now expected to be associated with the application using them. When upgrading from SES 3 or 4, the cluster will attempt to associate existing pools to known applications (that is, CephFS, RBD, and RGW).

In-use pools that are not associated with an application will generate a health warning. Unassociated pools can be manually associated using the new Ceph OSD pool application enable command.

The maximum number of Placement Groups per OSD before the monitor issues a warning has been reduced from 300 to 200 PGs. 200 is still twice the generally recommended target of 100 PGs per OSD. This limit can be adjusted via the mon_max_pg_per_osd option on the monitors. The older mon_pg_warn_max_per_osd option has been removed.

Creating pools or adjusting pg_num will now fail if the change would make the number of PGs per OSD exceed the configured mon_max_pg_per_osd limit. If it is necessary to create a pool with more PGs, this option can be adjusted.

RGW metadata search backed by ElasticSearch now supports end user requests service via RGW itself, and also supports custom metadata fields. A set of RESTful APIs was created, so users can search objects by their metadata. New APIs that allow control of custom metadata fields were also added.

RGW now supports dynamic bucket index sharding. This is now enabled by default: RGW will now automatically reshard the bucket index when the index grows beyond rgw_max_objs_per_shard. Hence, as the number of objects in a bucket grows, RGW will automatically reshard the bucket index in response. No user intervention or bucket size capacity planning is required.

RGW introduces server-side encryption of uploaded objects with three options for the management of encryption keys:

S3 Object Tagging API has been added: APIs for GET/PUT/DELETE object tags and the PUT object API are supported. However, there is no support for tags on Policies and Lifecycle yet.

RGW multisite now supports for enabling or disabling sync at a bucket level.

RGW now supports the S3 multipart object copy-part API.

You can now reshard an existing RGW bucket offline. Offline bucket resharding currently requires that all I/O (especially writes) to the specific bucket is quiesced.

RGW now supports data compression for objects.

RGW: the embedded Civetweb has been upgraded to version 1.8.

RGW: The Swift static website API is now supported (S3 support has been added previously).

RGW: S3 bucket lifecycle API has been added. Note that it currently only supports object expiration.

RGW: Support for custom search filters has been added to the LDAP authentication implementation.

Support for NFS version 3 has been added to the RGW NFS gateway.

A Python binding has been created for librgw.

RBD now has full, stable support for erasure coded pools via the new --data-pool option to rbd create.

RBD mirroring’s rbd-mirror daemon is now highly available. For reliability, SUSE recommends deploying several instances of rbd-mirror.

The name of the default pool used by the rbd CLI when no pool is specified can be overridden using the rbd default pool = POOL_NAME option.

Initial support for deferred image deletion via new rbd trash CLI commands. Images, even ones actively in-use by clones, can be moved to the trash and deleted at a later time.

New pool-level rbd mirror pool promote and rbd mirror pool demote commands to batch promote/demote all mirrored images within a pool.

Mirroring now optionally supports a configurable replication delay via the rbd mirroring replay delay = SECONDS configuration option.

Improved discard handling when the object map feature is enabled.

rbd CLI import and copy commands now detect sparse and preserve sparse regions.

RBD Images and Snapshots will now include a creation timestamp.

Specifying user authorization capabilities for RBD clients has been simplified. The general syntax for using RBD capability profiles is:

mon 'profile rbd' osd 'profile rbd[-read-only][pool=POOL-NAME[, …]]'

The rbd-mirror daemon now supports replicating dynamic image feature updates and image metadata key/value pairs from the primary image to the non-primary image.

The number of RBD image snapshots can be optionally restricted to a configurable maximum.

The RBD Python API now supports asynchronous IO operations.

CLI: the ceph -s/ceph status command has a fresh look.

CLI: ceph mgr metadata will dump metadata associated with each ceph-mgr daemon.

CLI: ceph versions/ceph osd,mds,mon,mgr versions summarize the version numbers of running daemons.

CLI: ceph osd,mds,mon,mgr count-metadata PROPERTY tabulates any other daemon metadata visible via the ceph osd,mds,mon,mgr metadata commands.

CLI: ceph features summarizes features and releases of connected clients and daemons.

CLI: ceph osd require-osd-release RELEASE replaces the old require_RELEASE_osds flags.

CLI: ceph osd pg-upmap, ceph osd rm-pg-upmap, ceph osd pg-upmap-items, ceph osd rm-pg-upmap-items can explicitly manage upmap items.

CLI: ceph osd getcrushmap returns a CRUSH map version number on stderr, and ceph osd setcrushmap VERSION will only inject an updated CRUSH map if the version matches.

This allows CRUSH maps to be updated offline and then reinjected into the cluster without fear of clobbering racing changes. For example, by newly added OSDs or changes by other administrators.

CLI: ceph osd create has been replaced by ceph osd new. This should be hidden from most users by user-facing tools like ceph-disk and DeepSea.

CLI: ceph osd destroy will mark an OSD destroyed and remove its cephx and lockbox keys. However, the OSD ID and CRUSH map entry will remain in place. This allows reusing the ID by a replacement device with minimal data rebalancing.

CLI: ceph osd purge will remove all traces of an OSD from the cluster, including its cephx encryption keys, dm-crypt lockbox keys, OSD ID, and CRUSH map entry.

CLI: ceph osd ls-tree NAME will output a list of OSD IDs under the given CRUSH name (like a host or rack name). This is useful for applying changes to entire subtrees. For example:

ceph osd down $(ceph osd ls-tree rack1)

CLI: ceph osd add,rm-noout,noin,nodown,noup allow applying the noout, noin, nodown, and noup flags to specific OSDs.

CLI: ceph osd safe-to-destroy OSD(s) will report whether it is safe to remove or destroy OSD(s) without reducing data durability or redundancy.

CLI: ceph osd ok-to-stop OSD(s) will report whether it is okay to stop OSD(s) without immediately compromising availability (that is, all PGs should remain active but may be degraded).

CLI: ceph log last N will output the last N lines of the cluster log.

CLI: ceph mgr dump will dump the MgrMap, including the currently active ceph-mgr daemon and any standbys.

CLI: ceph mgr module ls will list active ceph-mgr modules.

CLI: ceph mgr module enable,disable NAME will enable or disable the named ceph-mgr module. The module must be present in the configured mgr_module_path on the host(s) where ceph-mgr is running.

CLI: ceph osd crush ls NODE will list items (OSDs or other CRUSH nodes) directly beneath a given CRUSH node.

CLI: ceph osd crush swap-bucket SRC DEST will swap the contents of two CRUSH buckets in the hierarchy while preserving the bucket's IDs. This allows an entire subtree of devices to be replaced without disrupting the distribution of data across neighboring devices. For example, this can be used when replacing an entire host of FileStore OSDs with newly-imaged BlueStore OSDs.

CLI: ceph osd set-require-min-compat-client RELEASE configures the oldest client release the cluster is required to support. Other changes, like CRUSH tunables, will fail with an error if they would violate this setting. Changing this setting also fails if clients older than the specified release are currently connected to the cluster.

CLI: ceph config-key dump dumps config-key entries and their contents. The existing ceph config-key list only dumps the key names, not the values.

CLI: ceph config-key list is deprecated in favor of ceph config-key ls.

CLI: ceph config-key put is deprecated in favor of ceph config-key set.

CLI: ceph auth list is deprecated in favor of ceph auth ls.

CLI: ceph osd crush rule list is deprecated in favor of ceph osd crush rule ls.

CLI: ceph osd set-full,nearfull,backfillfull-ratio sets the cluster-wide ratio for various full thresholds: Respectively, when the cluster refuses I/O, when the cluster warns about being close to full, and when an OSD will defer rebalancing a PG to itself.

CLI: ceph osd reweightn will specify the reweight values for multiple OSDs in a single command. This is equivalent to a series of ceph osd reweight commands.

CLI: ceph osd crush set,rm-device-class manage the new CRUSH device class feature. Manually creating or deleting a device class name is generally not necessary as it will be smart enough to be self-managed.

ceph osd crush class ls and ceph osd crush class ls-osd will output all existing device classes and a list of OSD IDs under the given device class respectively.

CLI: ceph osd crush rule create-replicated replaces the ceph osd crush rule create-simple command to create a CRUSH rule for a replicated pool. It takes a class argument for the device class the rule should target (for example, hard disk or SSD).

CLI: ceph tell DAEMON help will now return a usage summary.

CLI: ceph fs authorize creates a new client key with caps automatically set to access the given CephFS file system.

CLI: The ceph health structured output (JSON or XML) no longer contains timechecks section describing the time sync status. This information is now available via the ceph time-sync-status command.

CLI: Certain extra fields in the ceph health structured output that used to appear if the monitors were low on disk space (which duplicated the information in the normal health warning messages) are now gone.

CLI: The ceph -w output no longer contains audit log entries by default. Add a --watch-channel=audit or --watch-channel=* to see them.

New ceph -w behavior: The ceph -w output no longer contains I/O rates, available space, PG info, etc. because these are no longer logged to the central log (which is what ceph -w shows). The same information can be obtained by running ceph pg stat.

Alternatively, I/O rates per pool can be determined using ceph osd pool stats. These commands do not self-update like ceph -w did. However, they have the ability to return formatted output by providing a --format=FORMAT option.

CLI: Added new commands pg force-recovery and pg force-backfill. Use them to boost recovery or backfill priority of specified PGs, so they are recovered/backfilled before any other. These commands do not interrupt ongoing recovery/backfill, but merely queue specified PGs before others, so they are recovered/backfilled as soon as possible.

The new commands pg cancel-force-recovery and pg cancel-force-backfill restore default recovery/backfill priority of previously forced PGs.

The osd crush location configuration option is no longer supported. Update your ceph.conf to use the crush location configuration option instead. To avoid movement of OSDs from a custom location to the default one, make sure to update your configuration file.

OSDs now avoid starting new scrubs while recovery is in progress. To revert to the old behavior in which the recovery activity does not affect scrub scheduling, set the following option:

osd scrub during recovery = true

The list of monitor hosts/addresses for building the monmap can now be obtained from DNS SRV records. The service name used when querying the DNS is defined in the mon_dns_srv_name configuration option which defaults to ceph-mon.

The osd class load list configuration option is a list of object class names that the OSD is permitted to load (or * for all classes). By default, it contains all existing in-tree classes for backward compatibility.

The osd class default list configuration option is a list of object class names (or * for all classes) that can be invoked by clients that have only capabilities *, x, class-read, or class-write. For backward compatibility, by default, it contains all existing in-tree classes. Invoking classes not listed in osd class default list requires a capability naming the class. For example, allow class foo.

The rgw rest getusage op compat configuration option allows dumping (or not dumping) the description of user statistics in the S3 GetUsage API. This option defaults to false. If the value is true, the reponse data for GetUsage looks like:

"stats": {
  "TotalBytes": 516,
  "TotalBytesRounded": 1024,
  "TotalEntries": 1
}

If the value is false, the reponse for GetUsage looks as it did before:

{
  516,
  1024,
  1
}

The osd out and osd in commands now preserve the OSD weight. That is, after marking an OSD out and then in, the weight will be the same as before (instead of being reset to 1.0). Previously, the monitors would only preserve the weight if the monitor automatically marked an OSD out and then in, but not when an administrator did so explicitly.

The ceph osd perf command will display commit_latency(ms) and apply_latency(ms). Previously, the names of these two columns were fs_commit_latency(ms) and fs_apply_latency(ms). The prefix fs_ has been removed because these values are not specific to FileStore.

Monitors will no longer allow pools to be removed by default. Monitors will only allow pools to be removed, when the setting mon_allow_pool_delete is true (defaults to false). This is an additional safeguard against pools being removed by accident.

If you have manually specified the monitor user rocksdb via the mon keyvaluedb = rocksdb option, you will need to manually add a file to the monitor's data directory to preserve this option:

echo rocksdb > /var/lib/ceph/mon/ceph-HOSTNAME/kv_backend

New monitors will now use rocksdb by default, but if that file is not present, existing monitors will use leveldb. The mon keyvaluedb option now only affects the back-end chosen when a monitor is created.

The osd crush initial weight option allows specifying a CRUSH weight for a newly added OSDs. However, the meaning of value 0 has changed:

If your configuration file explicitly specifies a value of 0 for this option, change it to a negative value (for example, -1) to preserve the current behavior.

The jerasure and shec plug-ins can now detect the SIMD instruction set at runtime and no longer need to be explicitly configured for different processors. Instead of processor-specific plug-ins, use jerasure or shec instead. The following plugins are now deprecated:

Using any of these plug-ins directly will create a warning in the monitor's log file.

Calculation of recovery priorities has been updated. Previously, this could lead to unintuitive recovery prioritization during cluster upgrades. In case of such a recovery: old-version OSDs would operate on different priority ranges than new ones. After the upgrade, the cluster will operate on consistent values.

The configuration option osd pool erasure code stripe width has been replaced by osd pool erasure code stripe unit, and given the ability to be overridden by the erasure code profile setting stripe_unit. For more information, see http://docs.ceph.com/docs/master/rados/operations/erasure-code-profile.

RBD and CephFS can use erasure coding with BlueStore. This can be enabled for a pool by setting allow_ec_overwrites to true.

This relies on checksumming of BlueStore to do deep scrubbing. Therefore, enabling this on a pool stored on FileStore is not allowed.

The JSON output of rados df now prints numeric values as numbers instead of strings.

The mon_osd_max_op_age option has been renamed to mon_osd_warn_op_age (default: 32 seconds), to indicate that a warning is generated at this age.

There is also a new option that controls when an error is generated: The option mon_osd_err_op_age_ratio is a expressed as a multiple of mon_osd_warn_op_age (default: 128, or around 60 minutes).

The default maximum size for a single RADOS object has been reduced from 100 GB to 128 MB. The 100 GB limit was completely impractical in practice while the 128 MB limit is high but reasonable.

If you have an application written directly to librados that is using objects larger than 128 MB, you can adjust the default value with osd_max_object_size.

Whiteout objects are objects which logically do not exist and which will return ENOENT if you try to access them. Previously, whiteout objects only occurred in cache tier pools. With SUSE Enterprise Storage 5 (Ceph Luminous), logically deleted but snapshot objects now also result in a whiteout object.

Whiteout objects can lead to confusion, because they are listed by rados ls and in object listings from librados but are otherwise inaccessible. To determine which objects are actually snapshots, use rados listsnaps.

The deprecated crush_ruleset property has been removed. For the commands osd pool get and osd pool set, use crush_rule instead.

The option osd pool default crush replicated ruleset has been removed and replaced by the osd pool default crush rule option. By default, it is set to -1. This means, the monitor will pick the first type of replicated rule in the CRUSH map for replicated pools.

Erasure-coded pools have rules that are automatically created for them if they are not specified when creating the pool.

SUSE recommends against using Btrfs with FileStore. To upgrade to SUSE Enterprise Storage 5 (Ceph luminous), If you are using Btrfs-based OSDs and want to upgrade to luminous, add the following to your ceph.conf:

enable experimental unrecoverable data corrupting features = btrfs

SUSE recommends moving these OSDs to FileStore with XFS or BlueStore.

The ruleset-* properties for the erasure code profiles have been renamed to crush-*. This was done to move away from the obsolete term ruleset and to be more clear about their purpose. There is also a new optional crush-device-class property to specify a CRUSH device class to use for the erasure-coded pool.

Existing erasure-code profiles will be converted automatically when the upgrade completes (that is, when ceph osd require-osd-release luminous is run). However, provisioning tools that create erasure-coded pools need to be updated.

The structure of the XML output for osd crush tree has changed slightly to better match the output of osd tree. The top-level structure is now nodes instead of crush_map_roots.

When assigning a network to the public network and not to the cluster network, the network specification of the public network will be used for the cluster network as well.

Previously, this would lead to cluster services being bound to 0.0.0.0:PORT, thus making the cluster service even more publicly available than the public services. When only specifying a cluster network, the public services will still bind to 0.0.0.0.

Previously, if a client sent an operation to the wrong OSD, the OSD would reply with ENXIO. The rationale here is that the client or OSD is buggy and the error should be made clear.

With SUSE Enterprise Storage 5 (Ceph Luminous), the reply ENXIO will only be sent if the option osd_enxio_on_misdirected_op option is enabled (off by default). This means that a VM using librbd that previously would have received an EIO and gone read-only will now see a blocked/hung IO instead.

The configuration option journaler allow split entries has been removed.

The configuration option mon_warn_osd_usage_min_max_delta has been removed and the associated health warning has been disabled. This option did not take into account clusters undergoing recovery or CRUSH rules that do not target all devices in the cluster.

A new configuration option public bind addr was added to support dynamic environments like Kubernetes. When set, the Ceph MON daemon could bind locally to an IP address and advertise a different IP address (public addr) on the network.

In CephFS, using multiple active MDS daemons is now considered stable. The number of active MDS servers can be adjusted up or down on an active CephFS file system.

CephFS directory fragmentation is now stable and enabled by default on new file systems. To enable it on existing file systems use ceph fs set FS_NAME allow_dirfrags.

Large or very busy directories are sharded and (potentially) distributed across multiple MDS daemons automatically.

CephFS directory subtrees can be explicitly pinned to specific MDS daemons in cases where the automatic load balancing is not desired or effective.

Client keys can now be created using the new ceph fs authorize command to create keys with access to the given CephFS file system and all of its data pools.

libcephfs function definitions have been changed to enable proper UID/GID control. The library version has been increased to reflect the interface change.

CephFS: Standby replay MDS daemons now consume less memory on work loads doing deletions.

CephFS: scrub now repairs backtrace, and populates damage ls with discovered errors.

CephFS: A new pg_files subcommand to cephfs-data-scan can identify files affected by a damaged or lost RADOS PG.

CephFS: False-positive failing to respond to cache pressure warnings have been fixed.

CephFS: Limiting MDS cache via a memory limit is now supported using the new mds_cache_memory_limit configuration option (1 GB by default). A cache reservation can also be specified using mds_cache_reservation as a percentage of the limit (5 % by default).

Limits by inode count are still supported using mds_cache_size. Setting mds_cache_size to 0 (the default) disables the inode limit.

When configuring ceph-fuse mounts in /etc/fstab, you can use the new syntax ceph.ARGUMENT=VALUE in the options column instead of putting configuration in the device column. However, the previous syntax style still works.

CephFS clients without the p flag in their authentication capability string can no longer set quotas or layout fields. Previously, this flag only restricted modification of the pool and namespace fields in layouts.

CephFS will generate a health warning if you have fewer standby daemons than it thinks you should have. By default, this will be 1 if you ever had a standby, and 0 if you did not. You can customize this using ceph fs set FS standby_count_wanted NUMBER. Setting it to 0 will disable the health check.

The ceph mds tell command has been removed. It has been superseded by ceph tell mds.ID.

The apply mode of cephfs-journal-tool has been removed.