StorPool 18.01 Release Change Log¶
18.01.267.08df121 – -> 18.01.328.4ab9a5f¶
Native 4k drives support
Various improvements in internal tools, monitoring and services
Changes in the way data on the underlying drives is aggregated in a more efficient manner
18.01.242.b4fbf45 – -> 18.01.267.08df121¶
No longer ignore snapshots being deleted when re-basing long snapshot-volume chains
Fix a bug in handling aborted iSCSI operations in
storpool_iscsi
serviceFix in
storpool_server
to not forget to delete moved snapshot objectsAdd support for centos 7.5 in the
storpool_rdma
kernel moduleAdd support for 4.14+ & OFED in the
storpool_rdma
kernel moduleAdd support for 4.12+ in the
storpool_pci
kernel moduleImprove some of the diagnostic and monitoring tools
Make the “recovery” state of a task clearly visible in the output of
storpool task list
18.01.228.60b8f69 – -> 18.01.242.b4fbf45¶
Disable check of BBU status for genreport
Fix service name and symlinks on CentOS 6 for the storpool_controller service
Fix installing dependency packages on Ubuntu 14.04
Fix the installation of the CentOS 6 cgmove init script
Remove “reload” for init scripts, as it’s not supported by the StorPool services and actually does “restart”
Add
/usr/lib/storpool/get_tech_support
- Collects useful information to be sent to support for initial troubleshooting in case of no access to the cluster.
18.01.181.cec460e – -> 18.01.228.60b8f69¶
Let the StorPool CLI prompt be customized using the SP_CLI_PROMPT variable
Use port 443 (instead of 8086) in the storpool_stat service to get through firewalls more easily
Add support for GCC 7
Remove log entries for recovered objects (prevents syslog flooding)
Multiple fixes and improvements on the server and bridge services
Handle the new fio 3.1+ log format in the disk_tester tool
Improve the fio test scripts
Run the mgmt and bridge services in a separate memory cgroup - mgmt.slice
Improve the storpool_bd kernel module
Add the SP_CPU_DMA_LATENCY variable to the
storpool.conf.example
file (more info at https://access.redhat.com/articles/65410 and https://www.kernel.org/doc/Documentation/power/pm_qos_interface.txt)Add a check for the status of the BBU/Cache Vault of RAID controllers
Improve repsync service
Add a tool for filtering git logs
Add a tool for rotating snapshots on number of volumes
18.01.91.edf2ec3 – -> 18.01.181.cec460e¶
Bridge redundancy, now multiple bridges may run at the same time with only one active at a given time
Bridge lower CPU when recovering objects with very small amount of actual data
Improve the site-to-site throughput on higher-latency connections
Support 4.15+ kernels and gcc 7
Let the configuration in
/etc/storpool.conf.d/*.conf
override the one in/etc/storpool.conf
Install the storpool_stat statistics collection tool
CLI is now showing the much more readable ‘RECOVERY’ (instead of 0) and ‘bridge’ (instead of 1) in the
task id
column in the output ofstorpool task list
Added a configuration option to skip the dependency installation
Various reliability fixes and improvements
18.01.41.eacd2bf – -> 18.01.91.edf2ec3¶
Various small fixes and improvements
Only execute the storpool_cgmove tool on boot
Remove an unused/obsolete service (machine-cgsetup)
Add example cgroup configuration for Debian 9
Install modules for all installed kernels on the system and for the latest available one in the package
Install genreport on all nodes and add non-API related checks
Add support for Intel P4500
Add the
/usr/lib/storpool/balancer.sh
toolBump the maximum number of volumes+snapshots to 30000
18.01.2.d5c0646 – -> 18.01.41.eacd2bf¶
Keep and update the balancer state through transactions
Various improvements and optimizations in the re-balancing tools
Add a flag to distinguish snapshots not being deleted due to multiple children
Fix storpool_bridge to not send older data and TRIMs
The storpool net list CLI output now includes the currently expected number of voting beacons
16.02.198.ee35bd2 – -> 18.01.2.d5c0646¶
Decrease boot time for storpool_server
New storpool_controller service to handle statistics data collection from the storpool_server processes in order to be used by the API
Remove all the empty objects for all the snapshots that are not attached, thus decreasing the number of objects used with many snapshots
The
placementGroup addDisk
command now only accepts diskIDs for disks already recognized by the clusterDisplay the servers that were up but have been down since the API was started
Support for 4.13 and 4.14 kernels
Various robustness and reliability fixes
16.02.168.df45b80 – -> 16.02.198.ee35bd2¶
Fix for storpool_block to properly handle StorPool volumes mounted with the _netdev option on shutdown
Fix for disk_tester when the drives are handled by the Dell PERC controller
Fix a harmless Perl interpreter warning in the StorPool CLI
A few other small fixes in tools and documentation
16.02-164.1829075 – -> 16.02.168.df45b80¶
Added simple short name/value tags stored in the volume/snapshot name both at creation time and afterwards
Added NVMe TRIM support
16.02.85.b833cc3 – -> 16.02-164.1829075¶
Fix an internal statistics script’s handling of the StorPool interface configuration
Fix in the dependency resolution when hwloc-nox is installed
Fix in dependency installation on Debian-like systems (no-install-recommends)
Adds support for the CentOS 7.4 kernel line to storpool_rdma
Various robustness and reliability fixes
Set the correct search path in a cron definition
Allow the system administrator to override the exclusion of the Mellanox drivers in the initrd to avoid some rare hangs at boot time
Run “sync” after the StorPool installation to make it much more probable that the files and directories have actually been created and written to the system disks
Install a much improved test infrastructure using the fio tool into
/usr/share/doc/storpool/examples/fio
Add the
deleteAfter
keyword to thesnapshot remote unexport
CLI command
16.02.11.eabf888 – -> 16.02.85.b833cc3¶
Various small improvements in the build process and the update tools
Peers with version < 16.02 will no longer be allowed to join the cluster
Fix: beacons won’t block with heavily loaded API
16.02.3.b638454 – -> 16.02.11.eabf888¶
Add the
reuseServer
flag for volumes, snapshots, and templates to allow placing more than one disk in a replication chain onto the same serverChecks and reports available kernels without modules for running StorPool services
Fix for setirqaff to properly setup interrupts on storage only nodes with missing machine.slice cgroup setup