The Ceph Blog

Ceph blog stories provide high-level spotlights on our customers all over the world

  • October 25, 2017
    New in Luminous: PG overdose protection

    Choosing the right number of PGs (“placement groups”) for your cluster is a bit of black art–and a usability nightmare.  Getting a reasonable value can have a big impact on a cluster’s performance and reliability, both in a good way and a bad way.  Unfortunately, over the past few years we’ve seen our share of …Read more

  • October 23, 2017
    New in Luminous: Zabbix

    The Ceph manager service (ceph-mgr) was introduced in the Kraken release, and in Luminous it has been extended with a number of new python modules. One of these is a module exporting overall cluster status and performance to Zabbix. Enabling the dashboard module The Zabbix module is included in the ceph-mgr package, so if you’ve …Read more

  • October 19, 2017
    New in Luminous: RADOS improvements

    RADOS is the reliable autonomic distributed object store that underpins Ceph, providing a reliable, highly available, and scalable storage service to other components.  As with every Ceph release, Luminous includes a range of improvements to the RADOS core code (mostly in the OSD and monitor) that benefit all object, block, and file users. Parallel monitor …Read more

  • October 16, 2017
    New in Luminous: Erasure Coding for RBD and CephFS

    Luminous now fully supports overwrites for erasure coded (EC) RADOS pools, allowing RBD and CephFS (as well as RGW) to directly consume erasure coded pools.  This has the potential to dramatically reduce the overall cost per terabyte of Ceph systems since the usual 3x storage overhead of replication can be reduced to more like 1.2x …Read more

  • October 10, 2017
    New in Luminous: CephFS metadata server memory limits

    The Ceph file system uses a cluster of metadata servers to provide an authoritative cache for the CephFS metadata stored in RADOS. The most basic reason for this is to maintain a hot set of metadata in memory without talking to the metadata pool in RADOS. Another important reason is to allow clients to also …Read more

  • October 6, 2017
    v10.2.10 Jewel released

    This point release brings a number of important bugfixes in all major components of Ceph, we recommend all Jewel 10.2.x users to upgrade. Notable Changes¶ build/ops: Add fix subcommand to ceph-disk, fix SELinux denials, and speed up upgrade from non-SELinux enabled ceph to an SELinux enabled one (issue#20077, issue#20184, issue#19545, pr#14346, Boris Ranto) build/ops: deb: …Read more

  • October 2, 2017
    New in Luminous: CephFS subtree pinning

    The Ceph file system (CephFS) allows for portions of the file system tree to be carved up into subtrees which can be managed authoritatively by multiple MDS ranks. This empowers the cluster to scale performance with the size and usage of the file system by simply adding more MDS servers into the cluster. Where possible, …Read more

  • September 28, 2017
    New in Luminous: Dashboard

    The Ceph manager service (ceph-mgr) was introduced in the Kraken release, and in Luminous it has been extended with a number of new python modules. One of these is a simple monitoring web page, simply called `dashboard`. Enabling the dashboard module The dashboard module is included in the ceph-mgr package, so if you’ve upgraded to …Read more

  • September 28, 2017
    v12.2.1 Luminous released

    This is the first bugfix release of Luminous v12.2.x long term stable release series. It contains a range of bug fixes and a few features across CephFS, RBD & RGW. We recommend all the users of 12.2.x series update. Notable Changes¶ Dynamic resharding is now enabled by default for RGW, RGW will now automatically reshard …Read more

  • September 26, 2017
    New in Luminous: CRUSH device classes

    The flexibility of the CRUSH map in controlling data placement in Ceph is one of the system’s great strengths.  It is also one of the most painful and awkward parts of the cluster to manage.  Previously, any non-trivial data placement policy required manual editing of the CRUSH map, either to adjust the hierarchy or to …Read more

  • September 20, 2017
    New in Luminous: Improved Scalability

    CERN has been a long-time Ceph user and active community member, running one of the largest production OpenStack clouds backed by Ceph.  They are also using Ceph for other storage use-cases, backing a range of high energy physics experiments.  Overall scalability is understandably an area of interest. Big Bang I and II In 2015, CERN …Read more

  • September 20, 2017
    New in Luminous: Multiple Active Metadata Servers in CephFS

    The Ceph file system (CephFS) is the file storage solution for Ceph. Since the Jewel release it has been deemed stable in configurations using a single active metadata server (with one or more standbys for redundancy). Now in Luminous, multiple active metadata servers configurations are stable and ready for deployment! This allows the CephFS metadata …Read more