Ambedded Announce the Release of UniVirStor Manager Version 2.15.18 for Ceph Octopus23 Mar, 2021
Ambedded announces the release of UVS manager v2.15.18 for Ceph Octopus
We are delighted to announce the release of UVS manager v2.15.18. v2.15.18 is our first release for Ceph Octopus. The new shipments of Mars 400 will use this version as their default release.
We modified a significant portion of UVS codes to introduce the Ceph new management and orchestration tool in this release. The latest release enhances many user experiences on UVS manager. As the Ceph Octopus requires Python 3 support, we also upgrade the CentOS to version 8.
Deploying a new Ceph cluster or upgrade to v2.15.18 on Mars 400 will automatically run Ceph MON, OSD, and MDS in containers.
Using the earlier version, customers can keep using the Ceph Nautilus or upgrade it to Ceph Octopus. We also released a UVS version 2.14.17-3 to fix some bugs in the UVS manager and upgrade the Ceph to version 14.2.16 for customers who want to stay in the Ceph Nautilus. Please refer to the 2.14.7-3 release notes. Ambedded will continue the support of Ceph Nautilus on the Mars 400 appliance.
If your Mars 400 cluster runs Ceph Nautilus, we provide the automatic updater to upgrade your Ceph cluster to Octopus without interrupting your storage service. The updater upgrades the Ceph, the Linux kernel, and CentOS. The updater will help you update your cluster node by node, and the update will not interrupt your service.
What's New in Ceph Octopus
- All Ceph daemons are now running as containers.
- New Ceph dashboard
- PG autoscaling: You can allow the cluster to either make recommendations or automatically tune PGs based on how you will use your ceph cluster by enabling pg-autoscaling.
- Partial object recovery replaces log-based recovery to improve the speed of recovery.
- Snapshot trimming improvement eliminates metadata in the OSD map that (previously) would grow with cluster age.
- Bluestore performance improvement
- Support new RBD client cache policy write-around as the default setting. The write-around policy differs from the write-back policy in that it does not attempt to service read requests from the cache, unlike the write-back policy, and is, therefore, faster for high-performance write workloads.
- Snapshot-based RBD mirroring reduces the I/O overhead of journal-based snapshots. Snapshot-based mirroring supports kernel RBD that journal-based mirroring doesn't.
- RGW supports object lock for Write-Once-Read-Many (WORM) data protection.
What's new in UVS Manager v2.15.18
- Ceph Version: 15.2.x (Octopus)
- Operating System
○ Upgrade Linux to CentOS 8
○ Kernel upgrade to 5.4.100
- Deployment & Management:
○ Use cephadm and ceph orchestrator as the deployment and management tools
○ Modify the UVS manager to meet the requirement of deploying ceph on containers
○ Modify the UVS manager to adopt the change of Ceph Octopus.
○ Modify the console RESTORE function to accelerate the container deployment.
- Automatically deploy NTP Server:When deploying the new cluster, the UVS manager will automatically use the node as the default NTP server when deploying a new cluster. This new feature avoids users forgetting to configure the NTP server. Users can change the NTP server after cluster deployment.
- Network Configuration:
○ Add configuration option for separating the Ceph public and in-cluster network.
○ Replace the Linux bonding module with Open vSwitch SLB bonding to provide a better stable network.
- OSD:Add an option for deploy OSD to use the SATA HDD/SSD as the bluestore storage.
- MON:Users can deploy multiple Ceph monitors in parallel to shorten the deployment time.
- RADOS Gateway
○ Simplify the deployment of new RGW
○ Unify the stand-alone and multisite RGW. All new deployed RGW will be multi-site ready.
- Ceph File System:Simplify the deployment of the Ceph file system. Now users can create the CephFS, pools, and metadata server (MDS) on one user interface.
- Demo license:A demo license is embedded in the new Mars 400 shipments. This simplifies the deployment of a new cluster.
- Bug Fixed:
○ Fix bug of unable to control the disk LEDs. (Backport to v2.14.17-3)
○ Fix bug of binding /var/tmp & /tmp all to M.2 SSD /tmp (Backport to v2.14.17-3)
○ Fix wrong setting of CephFS max_mds after the UVS manager upgrade (Backport to v2.14.17-3)
○ Use log rotation to prevent the logs from over consume the storage.
Upgrade from UVS 2.14.17
We made a video to demonstrate the upgrade procedure. Please check the YouTube video below.
The steps to completely upgrade the software are:
- Upgrade the first Monitor node by node admin console.
- Upgrade the rest of the monitor nodes by UVS manager.
- Upgrade the OSD nodes.
- Upgrade MDS nodes if you have MDS nodes in use.
- Check the UVS software versions.
- Upgrade the external RADOS gateways.
- Upgrade the external iSCSI gateways.
- Finally, please upgrade all of your ceph clients.
Customers who want to update your Ceph cluster, please contact Ambedded to get the update files and upgrade guide.
Demonstrate how the upgrade procedure Mars 400 from UVS v2.14.17 to v 2.15.18.
Lembar Data Mars 400 Mars 400SES
Alat Ceph & lembar data penyimpanan SUSE Enterprise
Ambedded - Solusi penyimpanan Ceph profesional untuk pasar.
Terletak di Taiwan sejak 2013, Ambedded Technology Co., LTD. adalah penyedia solusi penyimpanan Ceph yang menawarkan perangkat penyimpanan yang ditentukan perangkat lunak berbasis ARM dengan penghematan CapEX 60%, konsumsi daya 70% lebih sedikit, dan keunggulan penskalaan 100%.
Untuk perancangan dan pembuatan penyimpanan yang ditentukan Perangkat Lunak, tim Ambedded memiliki pengalaman luas untuk membantu pelanggan mengadopsi solusi ceph ke dalam industri serbaguna, seperti Telecom, medis, militer, pusat data edge, dan penyimpanan perusahaan yang diperlukan HA.
Ambedded telah menawarkan solusi penyimpanan Ceph kepada pelanggan sejak 2013, baik teknologi canggih maupun pengalaman 20 tahun, Ambedded memastikan setiap permintaan pelanggan terpenuhi.