StorPool 19.2 release notes

Release 19.2, more details here.

19.2 revision 19.01.2173.0ca29830b

Released: 07 July 2021

Moves handling of the heavier queries (VolumesSpace, SnapshotsSpace) back to the storpool_server service for a significant speed-up in clusters with many volumes and snapshots.
Moves per host status reporting from storpool_abrtsync to storpool_stat for lower latency on per-node monitoring alerts.

19.2 revision 19.01.2156.b87fc49d8

Released: 29 June 2021

Adds fix for storpool_ctl not detecting storpool_server instances properly when adding/removing/changing disks with different number of instances.
Adds an automatic cleanup of storpool_initdisk cached data on a change with the tool or the storpool_nvmed service.
The storpool_stat submodule responsible for reporting service issues will now skip storpool_server instances with ejected disks.

19.2 revision 19.01.2137.71224a0ce

Released: 19 June 2021

Adds fix for storpool_kubcsi/iSCSI combination causing an abort in storpool_mgmt service.
Adds kernel version override in /usr/lib/storpool/getpackage
Adds more validations in upgrade helper tooling.

19.2 revision 19.01.2120.40c434b67

Released: 11 June 2021

Adds support for CentOS 8.4 up to the 4.18.0-305.3.1.el8.x86_64 kernel.
Adds fix for dummy interfaces under LACP bonds to be considered as no-carrier interfaces.
Removes the no longer used storpool_reaffirm service.

19.2 revision 19.01.2099.32074a501

Released: 5 June 2021

Adds initial support for Intel E810 network interface controllers (ice driver).
Support for Ubuntu 20.04 kernel 5.8

19.2 revision 19.01.2084.34da44f18

Released: 26 May 2021

Adds fixes for the hpssacli_helper tool:
- now handles output for HP Smart Array Gen 10 controllers.
- skips failed drives, parses older output
Adds installer fixes to execute depmod only for the installed kernels.

19.2 revision 19.01.2068.b4804fe04

Released: 7 May 2021

Adds fix for iSCSI compare-and-write operations when IOPS limits are applied.
Fix for BGP announcement withdrawal in routed iSCSI networks.

19.2 revision 19.01.2016.c4bc17cda

Released: 30 Apr 2021

Adds support for HGST Ultrastar SN100 and Micron 9100 Pro devices.
Adds json output to the storpool_ctl tool.

19.2 revision 19.01.2012.bc34dd830

Released: 21 Apr 2021

Adds support for Toshiba Cx5 NVMe line of drives.
The loader now invokes storpool_hugepages after the /opt directory is mounted on boot.
Improves some logging messages for better clarity in beacon, API, and bridge.

19.2 revision 19.01.1991.f5ec6de23

Released 13 Apr 2021

Handles small MTU in iSCSI checks for portal group and controller addresses.
Adds srcversion property for kernel modules under Debian.
Enables the storpool-contrib repository in CentOS by default.
Adds the storpool_ctl helper tool. For details, see Managing services with storpool_ctl.

The new tool provides an easy way to perform an action for all installed services on a StorPool node. An example action is to start, stop, restart them, or enable/disable them on boot. Other such actions might be added later so that they could benefit from the additional safety checks already added in the present tool.
New implementation of the hpssacli_helper tool that caches the controller output (workaround for HP Smart Array opening all disks in readwrite mode).

19.2 revision 19.01.1957.1a9a9bb68

Released 30 Mar 2021

Fixes for snapshotSpace object properties in Python bindings.
The dsnaprebase tool is now considering transient snapshots as deleted when they no longer save any space.

19.2 revision 19.01.1949.1aea86986

Released 26 Mar 2021

Adds fix in API for the snapshotSpace call when spaceUsed is a negative value.

19.2 revision 19.01.1946.0b0b05206

Released 20 Mar 2021

Adds fixes in iSCSI checks in storpool_stat
- Adds support for interfaces with lower MTU on the initiators
- Adds fix for abort when nodes are without access to the StorPool API.
- Adds caching for the last known addresses and MTU for each known initiator.
  
  The storpool_stat service will now track and will raise an alert for all unreachable initiator addresses in each portal group that have at least one export and have connected to some of the portals at least once. With this change the monitoring system will better detect obscure networking issues. For more information, see Monitoring metrics collected.

19.2 revision 19.01.1942.0464eda88

Released 16 Mar 2021

Adds CLI help readability changes and fixes some spelling errors.
Adds fix for storpool_beacon LACP networks and SP_IGNORE_RX_PORT=1 configuration.
Support for Kubernetes worker nodes to attach StorPool volumes via iSCSI
Not specifying an export now exports a target to all initiators for a portal group.

19.2 revision 19.01.1907.2389092f3

Released 9 Mar 2021

Adds fix for a false positive check in the disk_tester utility.

19.2 revision 19.01.1905.5a6c7e113

Released 8 Mar 2021

The update_* tools are now covering also the storpool_volumecare service.
Adds fix for wrong MAC address initialization for the storpool_beacon service after NIC reset.
Adds support for generating bridge interfaces with automation tools.
Adds fix for ejecting the last disk on a running storpool_server instance not flagged as ejected.
Includes various other improvements in internal tools and services

19.2 revision 19.01.1879.4679753db

Released: 1 Mar 2021

Reliability improvements in the storpool_mgmt (API) and the storpool_bridge services.
API performance improvements in clusters with many volumes and snapshots (>20k).
Includes volumecare version 1.20 fixes and updates (Change history).

19.2 revision 19.01.1816.16fa37c0d

Released: 15 Feb 2021

The unprivileged storpool account used for sending debug reports is now a system one.

19.2 revision 19.01.1813.f4697d8c2

Released: 08 Feb 2021

Adds ability to see the status of remotely connected clusters; for details, see Status of remote bridges.
Sometimes connectivity issues might lead to slower bridge throughput and to occasional errors, which could now be tracked. We are planning to add an additional dashboard in analytics.storpool.com for the statistics collected from the bridge in the future so that periods with lower throughput could be correlated to other events when investigating for connectivity problem or when measuring throughput performance between clusters, bottlenecks when moving workloads between sub-clusters in a multicluster and others.

The new functionality is available in the CLI and the API (of which the CLI is just a frontend). Now showing:
- The state of the remote bridge (idle, connected)
- The TCP state of the configured remote bridges:
  Connection time
  
  TCP errors
  
  Time of the last error
  
  Bytes Sent/Received
  
  Additional TCP info (available from API only)
- Number of exports from the remote side
- Number of sent exports from the local side
Initial support for Ubuntu 20.04 LTS
Includes improvements and fixes in internal tools, and services.

19.2 revision 19.01.1795.5b374e835

Released: 30 Jan 2021

Adds support for the vfio-pci driver for NVMe devices.

The storpool_nvmed service now may optionally handle NVMe drives through the vfio-pci driver. This will provide better NVMe drives compatibility on some platforms, as the vfio-pci driver is widely used.

The storpool_nvmed service could now optionally bind NVMe drives to the vfio-pci driver instead of the currently default storpool_pci one. This can be done by setting the SP_NVME_PCI_DRIVER=vfio-pci configuration option; for details, see NVMe SSD drives.

The main reason this is not the new default is that the iommu=pt option is required for both Intel and AMD CPUs, and the intel_iommu=on option additionally for Intel CPUs only on the kernel command line. Thus, this might require setting up some parameters in existing clusters.

Note

A new tool is available for auto-detecting the CPU and fixing the GRUB configuration accordingly at /usr/lib/storpool/enable_grub_iommu.

The vfio-pci driver will become the new default for newly validated devices with present production installations gradually moving towards it as well.
Adds support for Western Digital Ultrastar SN200 NVMe device.
Includes volumecare version 1.19 fixes and updates (Change history).
Includes small fixes and additions in internal services (service checks, storpool_abrtsync, etc.).
Fixes a problem with OpenStack components running both in a container and on the host system.
Includes fix in storpool_bd kernel module now properly showing statistics under CentOS 8 and Ubuntu’s 5.xx kernels

19.2 revision 19.01.1732.02297f62b

Released: 15 Jan 2021

Adds fixes for iscsi_checks in storpool_stat when network interfaces are missing.
In-server disk tester latency measurement fixes.
Adds fixes in tools for hugepages reservation (NVMe namespace detection)
Updates CentOS 8 related debugging tools.

19.2 revision 19.01.1720.8c71b2ec3

Released: 12 Jan 2021

Adds support for Western Digital Gold NVMe U.2 (WDS960G1D0D).
Additional per-host checks in the storpool_stat service now collects checksums for all StorPool services and related configuration files.
Adds fix for aggregation with large SATA SSD disks (7TB+).
Adds visible function symbols for easier function tracing (i.e., with perf top).
Adds small fixes in some externally visible helper tools (netbandwidth, disk_tester), internal tools and services.

19.2 revision 19.01.1689.6a0aa758b

Released: 08 Jan 2021 - A major internal build system overhaul. - Adds initial support for the Intel NICs handled by the ice driver - Adds support for Toshiba XD5

19.2 revision 19.01.1656.1d6d61d3f

Released: 22 Dec 2020

Multicluster backup locations are now supported in volumecare (Change history).
New policy in volumecare (keep-daily-split).
Includes improvements and and fixes in internal tools, core and services.
Creating a template without replication is now forbidden.
Monitoring now also tracks connectivity between iSCSI controllers and initiators.

Previous release

19.1 revision 19.01.1628.1b627d0