Professional Documents
Culture Documents
Ifs Pub Upgrade Planning and Process Guide en Us
Ifs Pub Upgrade Planning and Process Guide en Us
January 2023
Notes, cautions, and warnings
NOTE: A NOTE indicates important information that helps you make better use of your product.
CAUTION: A CAUTION indicates either potential damage to hardware or loss of data and tells you how to avoid
the problem.
WARNING: A WARNING indicates a potential for property damage, personal injury, or death.
© 2013 - 2023 Dell Inc. or its subsidiaries. All rights reserved. Dell Technologies, Dell, and other trademarks are trademarks of Dell Inc. or its
subsidiaries. Other trademarks may be trademarks of their respective owners.
Contents
Contents 3
Backup data........................................................................................................................................................................20
SyncIQ backup............................................................................................................................................................. 20
NDMP backup.............................................................................................................................................................. 20
Back up custom settings............................................................................................................................................ 21
Complete or stop jobs in progress................................................................................................................................. 21
Complete system jobs................................................................................................................................................ 22
Update Node Firmware Package...................................................................................................................................22
Update Drive Support Package..................................................................................................................................... 22
Configure IPMI ports........................................................................................................................................................ 22
Disable SupportAssist or Secure Remote Services (SRS)...................................................................................... 22
4 Contents
Chapter 6: Troubleshooting your upgrade....................................................................................37
Troubleshooting an upgrade........................................................................................................................................... 37
Contents 5
1
Introduction to this guide
Topics:
• About this guide
• Where to get help
NOTE: For upgrades from OneFS 8.1.1.x and earlier, please contact your account team.
Planning an upgrade 7
Review required documentation
Reviewing the documentation in this list helps you to understand the upgrade process and the impact the upgrade could have on
your workflow. Some links require a support login.
Required documentation
● Current PowerScale Software Releases
Check on current OneFS releases, release dates, and which version has the Target Code status.
● PowerScale OneFS Release Notes and Maintenance Release Notes
Read the OneFS release notes for each version between your current version and your target version for information about
new features, changes, resolved issues, and known issues.
NOTE: The new features and known issues for all supported OneFS versions have been compiled into this guide in the
following appendices: Features; Known Issues.
Other documentation
● OneFS Upgrade Process Troubleshooting guide
Review this troubleshooting guide if you encounter issues during your upgrade.
● OneFS Technical and Security Advisories
Determine whether any PowerScale Technical Advisories or Security Advisories have been issued for the version of OneFS
to which you are upgrading.
Parallel upgrades
Dell Technologies recommends using the parallel upgrade option when upgrading any size cluster running OneFS 8.2.2 and later.
Parallel upgrades require a smaller maintenance window than rolling upgrades, and do not require the interruption of service like
simultaneous upgrades.
A parallel upgrade installs the new operating system on a subset of nodes and restarts that subset of nodes at the same time.
Each subset of nodes attempts to make a reservation for their turn to upgrade until all nodes are upgraded. Node subsets and
reservations are based on diskpool and node availability.
During a parallel upgrade, node subsets that are not being upgraded remain online and can continue serving clients. However,
clients that are connected to a restarting node are disconnected and reconnected. How the client connection behaves when
a node is restarted depends on several factors including client type, client configuration (mount type, timeout settings), IP
allocation method, and how the client connected to the cluster. In OneFS 9.2.0.0 and later, client connection behavior is
managed by the disruption manager settings.
8 Planning an upgrade
Rolling upgrades
A rolling upgrade installs the new operating system and restarts each node individually in the OneFS cluster so that only one
node is offline at a time. A rolling upgrade takes longer to complete than a simultaneous upgrade. During a rolling upgrade, nodes
that are not currently being upgraded remain online and can continue serving clients. However, clients that are connected to
a restarting node are disconnected and reconnected. How the client connection behaves when a node is restarted depends
on several factors including client type, client configuration (mount type, timeout settings), IP allocation method, and how the
client connected to the cluster.
Simultaneous upgrades
A simultaneous upgrade installs the new operating system and restarts all nodes in the OneFS cluster at the same time.
Simultaneous upgrades are faster than rolling upgrades but require a temporary interruption of service during the upgrade
process. All client connections to the cluster must be terminated before completing the upgrade and data is inaccessible until
the installation of the new OneFS operating system is complete and the cluster is back online.
isi_for_array uname -r
Find your version of OneFS in the supported OneFS Upgrade Paths matrix and confirm which versions of OneFS are supported
upgrade paths.
Planning an upgrade 9
Upgrade Paths for supported versions of OneFS
The following table can be used to determine the supported upgrade path from the current version of OneFS on your cluster to
a target version of OneFS. The upgrade paths table displays the supported upgrade paths from major OneFS version to major
OneFS version, but does not include patch information.
Key:
● "O" : Rolling and Simultaneous upgrades available
● "=" : Parallel, Rolling, and Simultaneous upgrades available
● "x" : Unsupported upgrade path
NOTE: These supported upgrade paths do not guarantee bug fix and feature parity. For more information about using
patches to achieve bug fix and feature parity, contact your account team.
NOTE: If you are upgrading from OneFS 8.1.x and using Mellanox switches, it is recommended that you first upgrade to
OneFS 8.2.2 before upgrading to the OneFS 9.x family. For more information, see the Flexnet unable to assign IP addresses
KB article.
NOTE: If you are upgrading from an earlier version of OneFS, contact your account team for assistance.
10 Planning an upgrade
NOTE: If you have a performance-sensitive workload, PowerScale recommends that you consult your Sales Engineer for
assistance during the pre-upgrade planning process.
Upgrade The maintenance window should encompass the pre-upgrade, upgrade, and post-upgrade phases.
maintenance Estimate the time that it takes to run the upgrade considering cluster size and upgrade type (parallel,
window rolling, or simultaneous). Schedule time to inform users when the upgrade will take place and that client
connections might be slow, file access might be affected, and clients might be disconnected. A best
practice is to upgrade the cluster during an off-hours maintenance window.
Schedule time for node and drive health checks and replacement of bad hardware. Include time to update
configurations and settings that are not supported in the new version.
Estimate the time that it takes to back up your data, considering cluster size, number of files, types of
files, and file size. Also include time to collect information about the cluster such as status, logs, and
settings.
If performing a parallel or rolling upgrade, consider whether you will configure client connection drain
times, which will extend the required maintenance window, but lower the impact on client connections.
Build in time to let the upgrade jobs run to completion and to reestablish permissions and connections.
Schedule time or extend the maintenance window to accommodate post-upgrade tasks such as
reconfiguring custom settings, updating scripts to reflect command and functionality changes in the
upgrade version, and potential troubleshooting.
(Optional) If available, upgrading a test cluster with the same current version of OneFS before you upgrade your
Upgrade a test production cluster can expose issues that could slow down or prevent the upgrade of your production
cluster system.
After you upgrade a test cluster, verify that the cluster is operational and validate key workflows on the
test cluster by simulating how administrators, users, and applications interact with the system.
Planning an upgrade 11
3
Completing pre-upgrade tasks
Topics:
• Pre-upgrade process - Overview
• Collect cluster information
• Check cluster readiness
• Verify configurations and settings
• Download the OneFS installation bundle file
• Upgrade compatibility check utility
• On-Cluster Analysis tool
• HealthCheck tool
• Backup data
• Complete or stop jobs in progress
• Update Node Firmware Package
• Update Drive Support Package
• Configure IPMI ports
• Disable SupportAssist or Secure Remote Services (SRS)
Pre-upgrade - Summary
The following is a summary of steps to perform during the pre-upgrade phase:
1. Collect cluster information.
a. Collect cluster status.
b. Gather cluster logs.
c. Check cluster hardware health.
d. Check cluster available space.
2. Resolve events and errors.
3. Verify cluster configuration.
a. Preserve the Kerberos keytab file.
b. Install DataIQ or InsightIQ.
4. Download the OneFS installation file.
5. Run the upgrade compatibility check utility.
a. Reconfigure unsupported SMB settings.
6. Run the On-Cluster Analysis tool.
7. Run the HealthCheck tool.
8. Back-up cluster data.
9. Complete system jobs.
10. Update Node Firmware Package.
11. Update Drive Support Package.
12. Configure IPMI ports.
13. Disable SupportAssist.
isi_gather_info
To gather the log files in OneFS 9.1.0.0 and later, run the following command:
The files generated during the log gathering process are stored on the cluster in the /ifs/data/Isilon_Support/pkg
directory.
isi status -v
2. To check for drives that do not report a status of HEALTHY, L3, or JOURNAL.
NOTE: If a drive is degraded, do not continue with the upgrade until the issue is resolved.
4. If the cluster has an InfiniBand network, confirm whether a node has been assigned the OpenSM (subnet manager) main
role.
Confirm that the output displays only one node in the cluster as the main (opensm). The output should be similar to the
following:
NOTE:
If the command does not return output that includes a node in the cluster as the main (opensm), then another device
has assumed the OpenSM main role. In that case, confirm that cables from another cluster are not connected to this
cluster's primary or secondary InfiniBand switch, and then run the command again. For more information about this
OpenSM issue, see KB 192527.
NOTE: If you are using the Dell S4112F-ON, Dell S4148F-ON, Dell Z9100-ON, or Dell Z9264-ON as back-end network
switches, see DTA 190714 for a workaround to a known upgrade issue.
/ifs The /ifs directory cannot be If this directory is at or near the minimum available-
more than 90 percent capacity. space requirement, see the following resources for
steps to address the issue:
● Event ID 100010004, The cluster's /ifs partition
is near capacity.
/var The /var partition cannot be If this directory is at or near the minimum available-
more than 90 percent capacity. space requirement, see the following resources for
steps to address the issue:
● Knowledge base article 471789.
● Event ID 100010001, The /var partition is near
capacity.
/var/crash The /var/crash directory If this directory is at or near the minimum available-
cannot be more than 90 percent space requirement, see the following resources for
capacity. steps to address the issue:
● Knowledge base article 458364.
● Event ID 100010002, The /var/crash partition is
near capacity.
isi stat
● To confirm how much space is being used in each node pool, run the following command:
isi stat -p
● To confirm how much space is being used by critical directories on the cluster, run the following command:
The isi_for_array output is similar to the following for each node in the cluster:
● If the command returns any critical errors, check the log files in the following directories for more information:
○ /var/log
○ /var/log/messages
○ /var/crash
● NOTE: If any log files contain messages about a dynamic sector recovery (DSR) failure or a Data Integrity (IDI)
failure, contact PowerScale before you upgrade.
2. Cancel non-critical events before upgrading to prevent a recurrence of notifications that you know to be harmless.
If you have critical events that you are unable to resolve, contact support before upgrading.
</files>
</user_preserve>
NOTE: If you use the web administration interface to upgrade from OneFS 9.3.0.x or earlier, to OneFS 9.4.0.0 or later,
you must first change the file extension from .isi to .tar.gz. (example: change OneFS_v9.4.0.0_Install.isi to
OneFS_v9.4.0.0.Install.tar.gz)
gzip -d /ifs/data/<installation-file-name>
NOTE: Clusters running OneFS 9.4.0.0 and later are not required to unzip .tgz files.
7. (OneFS 9.3.0.0 and earlier) To unbundle the .tar file, run the following command where <installation-file-name> is the name
of the OneFS install tar file:
NOTE: Clusters running OneFS 9.4.0.0 and later are not required to unbundle .tar files.
8. (OneFS 9.3.0.0 and earlier) Validate the manifest file (example: OneFS_v9.3.0.0_Install.tar.gz). See the Checking Manifests
section of the PowerScale OneFS Security Configuration Guide for more information.
NOTE: Clusters running OneFS 9.4.0.0 and later use the OneFS Catalog to verify packages.
1. Open a secure shell (SSH) connection to any node in the cluster and log in to the cluster with the root account.
2. Start the upgrade compatibility check utility by running the following command, where <install-image-path> is the file path
of the upgrade installation file.
isi upgrade cluster assess <install-image-path>
HealthCheck tool
The HealthCheck tool enables you to evaluate the status of specific software and hardware components of your cluster and
your cluster environment. For more information about the HealthCheck tool, see this site: HealthCheck Info Hub
Backup data
It is recommended to backup your cluster data immediately before you upgrade. Schedule sufficient time for the back up to
complete before the upgrade window.
SyncIQ backup
SyncIQ is one option you can use to backup your OneFS cluster. SyncIQ creates and references snapshots to replicate a
consistent point-in-time image of a root directory.
For more information about backing up your OneFS cluster, see the OneFS CLI Administration Guide or the OneFS Web
Administration Guide for your version of OneFS.
NOTE: If you are upgrading your cluster from OneFS 8.1.0.x or earlier to OneFS 8.1.1.x, 8.1.2.x, or 8.1.3.x, and your cluster
is in Compliance mode, you must ensure that all SyncIQ partners are on the same code and patch level OneFS before
restarting SyncIQ backups, or the backups fail. This issue is resolved in OneFS 8.2.0 and later.
NOTE: If you are upgrading your cluster to OneFS 8.2.0 through 9.2.1.0, are using SyncIQ, and have non-networked
(NANON) nodes on your source cluster, SyncIQ jobs might fail after the upgrade. For more information including a
workaround, see the KB article SyncIQ jobs failing with NANON after upgrade
NDMP backup
Other OneFS cluster backup options include using the Network Data Management Protocol (NDMP).
From a backup server, you can perform both NDMP three-way backup and NDMP two-way backup processes between a cluster
and backup devices such as tape devices, media servers, and virtual tape libraries (VTLs).
See the OneFS Web Administration Guide or the OneFS CLI Administration Guide for information about backing up data using
NDMP.
Passwords for After you upgrade, you might Make a list of the local accounts and their passwords before you
local user have to reset the passwords upgrade.
accounts of the local user accounts that
you configured on the cluster.
Other users should be prepared to
reset the passwords of their local
accounts after the upgrade.
sysctl If you changed the default PowerScale does not recommend modifying sysctl parameters
parameters value assigned to one or unless you are instructed to do so by Dell Technologies Support. If
more sysctl parameters you must modify a sysctl parameter, configure the parameter in
by editing the /etc/mcp/ the /etc/mcp/override/sysctl.conf file to ensure that the
override/sysctl.conf file, change is preserved when you upgrade a node or a cluster.
the etc/mcp/templates
Before you upgrade, document your custom sysctl parameters and
directory, or the /etc/local/
back up the /etc/mcp/override/sysctl.conf, the etc/mcp/
sysctl.conf file, you might
templates directory, and the /etc/local/sysctl.conf files.
need to reset the parameter after
you upgrade. For more information about making sysctl changes persist through
upgrades, see KB article 000102543 and KB article 000083411.
If you modified a sysctl
parameter by editing another
file—for example, the /etc/
sysctl.conf file—the change
will not be preserved during the
upgrade.
Aspera You must reinstall and reconfigure Before you upgrade, back up the Aspera configuration files in the
Aspera after you upgrade. following directories:
NOTE: Aspera is no longer ● /ifs/.ifsvar/aspera/etc/
supported In OneFS 8.2.0 and ● /ifs/.ifsvar/aspera/www/
later. ● /usr/local/aspera/var/aspera-prepost
To determine which version of Aspera is compatible with the version
of OneFS to which you are upgrading, see the Supportability and
Compatibility Guide, and then go to the Aspera website to download
the Aspera install files.
Cron jobs Cron jobs settings that were Document and back up custom cron job settings or configure them
not configured in the /etc/mcp/ in the /etc/mcp/override/crontab.smbtime file before you
override/crontab.smbtime upgrade.
file are not preserved during an
upgrade. After you upgrade, you might have to modify a cron job to
accommodate changes to OneFS commands.
2. To cancel a job, run the following command where <job_id> is the ID of the job you want to cancel:
NOTE: Do not cancel the Upgrade, FlexProtect, FlexProtectLin, or IntegrityScan jobs. If any of these four system jobs
are running, you cannot continue with the upgrade. If an Upgrade, FlexProtect, FlexProtectLin, or IntegrityScan system
job takes longer than expected to complete, contact Dell Technologies Support.
NOTE: Sync polices and jobs must be canceled or paused in order for the upgrade to complete successfully.
NOTE: Use of IPMI ports is supported in OneFS version 8.2.2 and later.
NOTE: You can upgrade OneFS using the command-line interface or the web administration interface.
Upgrade - Summary
The following is a summary of steps to perform during the upgrade phase:
1. Perform the upgrade.
2. Commit the upgrade.
3. Verify the upgrade.
NOTE: In 9.2.0.0 and later, you can use the upgrade status screen to monitor the progress of your upgrade, view
connected clients, and delay draining clients from specific nodes.
After the upgrade, a number of upgrade-related jobs may continue to run on the cluster for some time. During this time, the
cluster is accessible, but you might experience a decrease in cluster performance. After the jobs complete, performance will
return to normal. At this stage, the upgrade is complete, but is not committed. You can still roll back to the previous version of
OneFS. Some new features in the upgrade might not be available until the upgrade is committed.
In OneFS 9.2.0.0 and later, you can include the firmware upgrade with your OneFS upgrade. To perform a parallel upgrade
that includes a firmware upgrade, run the following command where <install-image-path> is the file path of the upgrade
install image and <firmware-path> is the file path of the firmware package.
NOTE: The isi upgrade cluster command runs asynchronously. The command does not run the entire upgrade
process; instead, it sets up the upgrade process, which nodes take turns controlling. For this reason, the command
returns quickly. To view the progress of the upgrade, use the isi upgrade view command or the web administration
interface.
Options Description
Nodes to Upgrade specific nodes with the --nodes <integer_range_list> option.
select for
Specify the nodes in their upgrade order as a comma-separated list (for example, --nodes 7,3,2,5) or as
upgrade
a dash-separated range (for example, --nodes 1-7) of logical node numbers (LNNs).
NOTE: We recommend that you upgrade all the nodes. If you upgrade some nodes, a weekly alert is
sent to confirm that the upgrade is making progress. Do not leave the cluster in a partially upgraded
state for a prolonged period. Some new features in the upgrade might not be available until all the nodes
in the cluster have been upgraded and the upgrade is committed. Refer to the release notes for the
OneFS version that you are upgrading to for information about features that require all the nodes to be
upgraded.
OneFS 9.2.0.0 and later ignores the node restart order. Instead, OneFS restarts any node that is ready, as long as only one
node is restarting at a time.
The following example for OneFS 8.2.2 through OneFS 9.1.0.x starts a parallel upgrade on nodes 7,3,2,5, in that order:
6. To add any remaining nodes to the upgrade, run the following command:
After the upgrade, a number of upgrade-related jobs may continue to run on the cluster for some time. During this time, the
cluster is accessible, but you might experience a decrease in cluster performance. After the jobs complete, performance will
return to normal. At this stage, the upgrade is complete, but is not committed. You can still roll back to the previous version of
OneFS. Some new features in the upgrade might not be available until the upgrade is committed.
SMB3 Client transitions from the restarting node to a new node without disruption.
NFSv2 and NFSv3 Client transitions from the restarting node to a new node without disruption.
NFSv4 Client transitions from the restarting node to a new node without disruption. Clients use NFSv4
failover support.
NOTE: For more information about NFS settings, see article 457328, Best practices for NFS client settings.
NOTE: In 9.2.0.0 and later, you can use the upgrade status screen to monitor the progress of your upgrade and view
connected clients.
After the upgrade, a number of upgrade-related jobs may continue to run on the cluster for some time. During this time, the
cluster is accessible, but you might experience a decrease in cluster performance. After the jobs complete, performance will
return to normal. At this stage, the upgrade is complete, but is not committed. You can still roll back to the previous version of
OneFS. Some new features in the upgrade might not be available until the upgrade is committed.
NOTE: The isi upgrade cluster command runs asynchronously. The command does not run the entire upgrade
process; instead, it sets up the upgrade process, which nodes take turns controlling. For this reason, the command
returns quickly. To view the progress of the upgrade, use the isi upgrade view command or the web administration
interface.
3. Optional: You can specify the following rolling upgrade options:
Options Description
Nodes to Upgrade specific nodes with the --nodes <integer_range_list> option.
select for
Specify the nodes in their upgrade order as a comma-separated list (for example, --nodes 7,3,2,5) or as
upgrade
a dash-separated range (for example, --nodes 1-7) of logical node numbers (LNNs).
NOTE: We recommend that you upgrade all the nodes. If you upgrade some nodes, a weekly alert is
sent to confirm that the upgrade is making progress. Do not leave the cluster in a partially upgraded
state for a prolonged period. Some new features in the upgrade might not be available until all the nodes
in the cluster have been upgraded and the upgrade is committed. Refer to the release notes for the
OneFS version that you are upgrading to for information about features that require all the nodes to be
upgraded.
OneFS 9.2.0.0 and later ignores the node restart order. Instead, OneFS restarts any node that is ready, as long as only one
node is restarting at a time.
The following example for OneFS 8.2.2 through OneFS 9.1.0.x starts a rolling upgrade on nodes 7,3,2,5, in that order:
The following example for OneFS 8.2.1 and earlier starts a rolling upgrade on nodes 7,3,2,5 in that order:
4. Optional: To add nodes to a current upgrade, run the following command, where <nodes> is the list of LNNs:
5. Optional: To add any remaining nodes to the upgrade, run the following command:
After the upgrade, a number of upgrade-related jobs may continue to run on the cluster for some time. During this time, the
cluster is accessible, but you might experience a decrease in cluster performance. After the jobs complete, performance will
SMB3 Client transitions from the restarting node to a new node without disruption.
NFSv2 and NFSv3 Client transitions from the restarting node to a new node without disruption.
NFSv4 Client transitions from the restarting node to a new node without disruption. Clients use NFSv4
failover support.
NOTE: If a client is reconnected to a node that has not yet been upgraded, the client might be required to reestablish a
connection to the cluster more than once.
NOTE: For more information about NFS settings, see article 457328, Best practices for NFS client settings.
NOTE: The isi upgrade cluster command runs asynchronously. The command does not run the entire upgrade
process; instead, it sets up the upgrade process, which nodes take turns controlling. For this reason, the command
returns quickly. To view the progress of the upgrade, use the isi upgrade view command or the web administration
interface.
After the upgrade, a number of upgrade-related jobs may continue to run on the cluster for some time. During this time, the
cluster is accessible, but you might experience a decrease in cluster performance. After the jobs complete, performance will
return to normal. At this stage, the upgrade is complete, but is not committed. You can still roll back to the previous version of
OneFS. Some new features in the upgrade might not be available until the upgrade is committed.
isi stat
2. Remove the installation files from the /ifs/data directory by running the following command where
<installation_file_name> is the name of the installation file:
rm /ifs/data/<installation_file_name>
3. Collect information about the cluster in OneFS 9.0.0.0 and earlier, by running the following command:
isi_gather_info
Collect information about the cluster in OneFS 9.1.0.0 and later, by running the following command:
Post-upgrade - Summary
The following is a summary of steps to perform during the post-upgrade phase:
1. Allow upgrade jobs to run.
2. Verify operational status.
3. Re-establish user privileges.
4. Restore client connections and workflows.
5. Verify Kerberos migration.
6. Restore custom settings.
7. Test custom scripts.
8. (Optional) Install latest patch.
NOTE: For more information about the Upgrade job, see KB Article 194551.
isi_for_array -s uname -a
2. View the status of the cluster and ensure all the nodes are operational:
isi status
3. Check the devices in the nodes to validate the status of the drives:
6. To verify network connectivity and SmartConnect functionality, ping all internal and external interfaces on the cluster.
7. Verify the network interfaces:
10. To check for issues, review the log files on the cluster:
isi_upgrade_logs
cat /var/log/messages
13. Check the status of the node firmware to ensure it is consistent across nodes:
For OneFS 9.0.0 and later:
isi_upgrade_logs --get-fw-report
14. Ensure that all the licenses carried over and remain up to date:
15. Check the status of the authentication providers to ensure that they remain active:
19. If you use SupportAssist or SRS, confirm that the service is reenabled.
NOTE: If you are using NDMP backups on your cluster, re-enable the NDMP service and test that it's working correctly.
Reinstall Aspera
If you were running Aspera for PowerScale before you upgraded your cluster, you must reinstall Aspera after you upgrade.
Aspera is not supported in OneFS 8.2.0 and later.
1. Run the following command to verify that the upgrade process enabled the Aspera Central service and the Aspera node
daemon:
isi services
2. If the Aspera service is not running properly, you must download and re-install a compatible version of Aspera from the
Aspera Enterprise Server website and verify that the Aspera Central service is enabled.
● For information about which versions of Aspera are compatible with your version of OneFS, see the PowerScale
Supportability and Compatibility Guide.
Troubleshooting an upgrade
If you experience problems with your upgrade, try the following:
● Check the upgrade logs and review for errors.
● Read through the Customer Troubleshooting Guide: OneFS Upgrade Process Flowchart for information that is related to
upgrade failures and error messages.
● Search for OneFS upgrade information within the knowledge base on the PowerScale support site.
● Contact your Dell Technologies Support representative.
OneFS 9.5.0.0
Full release notes available at: https://dl.dell.com/content/manual58071479-powerscale-onefs-9-5-0-0-release-notes.pdf
New features
This section summarizes the new features in OneFS 9.5.0.0.
SupportAssist SupportAssist is the new remote connectivity system that connects your OneFS cluster with Dell
introduced as Technologies Support. For more information, see the SupportAssist section in the OneFS Administration
the new way guide.
to connect with
Dell Technologies
Support.
Host-based The firewall controls inbound traffic on the front-end network. The firewall comes with predefined default
firewall added to policies that protect the OneFS default ports. You can modify the default policies and create custom
OneFS. policies. New Web UI options and isi network firewall commands are available.
Account policy Security restrictions can be configured to enforce a delay for subsequent logins after a failed
restrictions have administrative login. You can also disable inactive local accounts after a specified number of days, and
been updated. limit the number of active sessions a user can have on a node.
Expanded Password complexity is configurable for each local provider, and password policies can be configured for
password policy each access zone.
configuration.
Support If the Active Directory domain requires encryption, the redirector will enable and use encryption.
for Active
Directory domain
encryption
added.
TLS options You can configure PKI communication between the OneFS cluster and an LDAP server to use TLS. Enable
added for LDAP TLS and configure options in the isi auth ldap create or isi auth ldap modify commands.
communications. Options define levels of verification during the TLS handshake and include revocation checking against
Online Certificate Status Protocol (OCSP) URIs.
NFS persisted The NFS locks and waiters functionality are updated in the Web UI and CLI commands with new listing
locks and waiters options.
listings updated.
Expanded The OneFS web interface now includes the ability to import and manage your certificates. For more
certificate information, see the Certificates section of the OneFS Web Administration Guide.
management in
the web
interface.
HealthCheck Removed the HealthCheck definition package list from the output of the patch CLI isi upgrade
definition patch list.
packages have
been enhanced.
OneFS PAPI Security command added to configure authorization and authenticity of peripherals (typically attached
command added using USB connections to the node's hardware).
to enable or
disable USB port
connectivity.
Command to You can disable all USB ports across the cluster with an option that was added to the isi security
disable all USB settings modify command. The STIG hardening profile disables all USB ports across the cluster.
peripheral ports
is added.
Configurable Set a maximum threshold limit that stops moving files from source to target node pools or tier.
option added
for setting a
threshold limit of
SmartPools.
SyncIQ updated SyncIQ target location files are updated to display the same create time (c-time) as the file in the source
file create time in location. This update takes effect as files are synced.
target location.
Support for the Support for the Nvidia Connectx-6 Ethernet adapter has been added to OneFS in version 9.5.0.0.
Nvidia ConnectX-
6 Ethernet
Adapter added.
Remove earlier Once a version of a software component (like Python or SQLite) is installed, OneFS removes any earlier
versions of versions of that software component.
software
components.
The Python The Python version has been updated from 2.7.x to 3.8.x in OneFS.
version has been
updated.
Expanded A new IPv6 configuration option enables or disables IPv6 on the cluster. Other options include several
support for IPv6 duplicate address detection (DAD) settings, IPv6 automatic configuration, local link generation, and
on OneFS. processing of ICMPv6 redirect messages. IPv6 configuration was changed to use the isi network
external modify, isi network pools create, and isi network pools modify
commands.
OneFS auditing System audits collect platform events, such as shutdowns and reboots, and account related events, such
includes a new as password changes.
system audit
topic.
Syslog The syslog forwarding feature in the auditing module forwards logs directly to the configured remote
forwarding in the servers. The local collection that occurred in previous releases is eliminated.
audit module is
improved.
TLS options You can configure syslog forwarding to use TLS. You can configure each audit topic (configuration,
added to syslog protocol, and system) separately for TLS. Each audit topic.
forwarding in the
audit module.
Root password The system audit topic captures root password changes. This feature is always on and cannot be disabled.
changes are
audited.
Restricted CLI The Restricted CLI is a new login shell. All commands that are issued in the Restricted CLI are logged.
provides an No file access is permitted. For information about the Restricted CLI, see the PowerScale OneFS 9.5.0.0
audited and Security Configuration Guide.
limited CLI.
The bash The bash shell is updated to version 5.1.16.
shell has been
updated.
Rekey feature You can reencrypt the OneFS keystores without interrupting processing. There are separate reencrypting
added to the key commands for the cluster key store and the self-encrypted drive (SED) key stores.
store manager.
Single sign-on The WebUI includes a wizard for enabling single sign-ons.
capability added
to the WebUI.
Apache changes The OneFS HTTP data paths and control paths are separated. The two paths use separate ports.
have been added. Configuration parameters in the isi http settings modify command can set Apache directives
that control session timeouts.
FIPS 140-2 The STIG hardening profile implements the FIPS 140-2 cryptographic algorithms and disables services
compliance that do not comply with the standard. In nonhardening mode, you can enable FIPS mode to use the
for hardening cryptographic algorithms.
clusters.
Improved The workflows for applying a hardening profile and returning to system defaults are simplified.
workflows in
the Hardening
Module.
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.4.0.0
Full release notes available at: https://dl.dell.com/content/manual62544979-powerscale-onefs-9-4-0-0-release-notes.pdf
New features
This section summarizes the new features in OneFS 9.4.0.0.
Support for QLC Support for QLC SED Non-FIPS drives has been added to OneFS in version 9.4.0.9.
SED Non-FIPS NOTE: Your cluster must be on OneFS 9.4.0.9 or later to access this feature.
drives
Performance OneFS 9.4.0.0 includes numerous performance enhancements.
enhancements
New Datamover OneFS 9.4.0.0 adds the Datamover data transfer engine, enabling you to transfer data between
data transfer PowerScale clusters and object-at-scale cloud systems. Datamover ensures that copies of your data are
engine consistent among PowerScale clusters and/or cloud platforms, and allows you to control the frequency of
data transfers at scheduled times using policies. You can transfer data at the directory level and exclude
specific files and subdirectories from being transferred.
The embedded Datamover feature provides data replication for file and object deployments on-premises
or in the cloud, enables file-to-file transfers between PowerScale clusters using RPC, and file-to-object
copy transfers to S3 (ECS, AWS) and Azure cloud systems. Datamover provides data protection, data
repurposing (copy), and data archiving, with flexible push/pull data transfers between systems. For
details, see the OneFS 9.4.0.0 CLI Administration Guide.
Rolling The rolling migration from InfiniBand to Ethernet begins with Generation 6 clusters in OneFS 9.4.0.0. You
InfiniBand should replace any InfiniBand switches with Ethernet switches. Contact your Dell representative for more
to Ethernet information.
migration
isi_gather_in The isi_gather_info HTTP upload method is deprecated in OneFS 9.4.0.0. The HTTP upload
fo: HTTP upload method will be removed entirely in a release later in 2022. The alternative upload methods are ESRS
method (preferred) and FTP.
deprecation In the OneFS release following 9.4.0.0, the names of the HTTP upload options will change to emphasize
that they are plain text:
● isi_gather_info: --http will be renamed to --http-insecure
● isi diagnostics gather: --http-upload will be renamed to --http-insecure-upload
Long filename OneFS 9.4.0.0 increases the maximum supported filename length to 1024 bytes. The maximum path
support length is increased to 4096 bytes.
Enhancements OneFS SmartConnect supports viewing status information for nodes in a network pool with the CLI
to SmartConnect command isi network pools status. See the OneFS 9.4.0.0 CLI Administration Guide and
diagnostics OneFS9.4.0.0 CLI Command Reference for details.
Disable A security improvement provides granularity for enabling or disabling HTTP services. The new isi
nonessential services http modify command enables or disables the following services individually: restful
HTTP services access namespace (RAN), PAPI , WebUI, and remote management API. For more information, see the
OneFS 9.4.0.0 Security Configuration Guide.
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.3.0.0
Full release notes available at: https://dl.dell.com/content/manual64485109
Enhancements OneFS 9.3.0.0 provides granular privileges. Each functional area of cluster administration is associated
to role-based with a top-level or parent privilege. Each parent privilege can have one or more subprivileges, which can
access control: also have subprivileges. Granular privileges enable you to control the specific actions that a user can
granular perform within a cluster administration area in a detailed way. Administrators can grant the following
privileges types of privileges:
● No permission (-): The user cannot access the functional area . (New)
● Read (r): The user can view the functional area, for example, viewing SyncIQ policies.
● Execute (x): The user can read and run jobs that are associated with the functional area. Create and
update permissions are allowed only for specific URIs. For example, execute permission is configured
to view and run SyncIQ policies. (New)
● Write (w): The user has full access to the functional area.
See the OneFS 9.3.0.0 CLI Administration Guide or OneFS 9.3.0.0 Web Administration Guide for privileges
details. See the OneFS API Reference Guide or the API Marketplace for API details.
Upgrade: OneFS supports transparent SMB node reboots during the upgrade process.
transparent SMB
disconnection
and reconnection
on node reboots
Support OneFS 9.3.0.0 and later releases support cluster-wide session management. A session cookie for a OneFS
for cluster API or Web UI session applies across all nodes in the cluster with no need to reconnect if the session
coherent session moves from one node to another. See the PowerScale OneFS 9.3.0.0 API Reference Guide for details.
management
Long filename 9.3.0.0 increases the maximum supported filename length to 1024 bytes. The maximum path length is
support increased to 4096 bytes.
Support for OneFS 9.3.0.0 adds support for HDFS ACL. Ranger ACL continues to be supported.
HDFS ACL
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.2.1.0
Full release notes available at: https://dl.dell.com/content/docu104030
New features
This section summarizes the new features in OneFS 9.2.1.0.
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.2.0.0
Full release notes available at: https://dl.dell.com/content/docu103229
New features
This section summarizes the new features in OneFS 9.2.0.0.
Support OneFS 9.2.0.0 supports backing up and exporting the current configuration of a PowerScale cluster
for cluster for test and auditing purposes. See the 9.2.0.0 PowerScale OneFS CLI Administration Guide or Web
configuration Administration Guide for details.
export/import
Permission In OneFS 9.2.0.0, the system default umask setting is changed to 077 and the hardening STIG profile
changes rules are greatly simplified. Applying and reverting the STIG profile is faster.
New RBAC The ISI_PRIV_RESTRICTED_AUTH privilege enables granting a subset of rights to allow an administrator
privilege: to run user and group management actions, such as finding and listing users, setting user passwords,
ISI_PRIV_RESTR unlocking accounts, and adding and removing users and groups. The ISI_PRIV_RESTRICTED_AUTH
ICTED_AUTH privilege is intended for help or service desk types of administrators who do not need the full
ISI_PRIV_AUTH set of privileges. Administrators with the ISI_PRIV_RESTRICTED_AUTH privilege cannot
modify access level privileges, such as root level access to a cluster. Administrators with the restricted
auth privilege can only modify users and groups that have the same or less privilege.
External key This enhancement enables you to offload the SED data encryption key to an external KMIP-compliant Key
management for Management Server. See the 9.2.0.0 PowerScale OneFS CLI Administration Guide or Web Administration
self-encrypted Guidefor details.
drives (SEDs)
FIPS compliance PowerScale OneFS replaced the FIPS 140-2 cryptographic module from OpenSSL with a private label
validation module that provides updated FIPS compliance.
OpenSSL update The OpenSSL version has been updated to OpenSSL 1.0.2u.
IPv6 compliance 9.2.0.0 PowerScale OneFS is IPv6/USGv6 compliant.
RDMA support The NFS over RDMA feature (NFSoRDMA) is supported for NFSv3. This enhancement adds another
for NFS front-end network transport communication mechanism between the client and OneFS node. See the
9.2.0.0 PowerScale OneFS CLI Administration Guide or Web Administration Guide for details.
S3 enhancement The S3 protocol on OneFS is enhanced with consistent ETAG functionality. Two new options, use-
md5-for-etag and validate-content-md5 were added. See the 9.2.0.0 PowerScale OneFS CLI
Administration Guide or Web Administration Guide for details.
HTTP basic HTTP basic authentication is disabled by default. To enable HTTP basic authentication from the WebUI:
authentication 1. Log in to the WebUI as a user with administrative privileges.
is disabled by 2. Click Protocols > HTTP settings.
default
3. In the Authentication settings section, click the Authentication drop-down and choose the HTTP
authentication type.
4. Save changes.
See the 9.2.0.0 PowerScale OneFS CLI Administration Guide or Web Administration Guide for details.
Improvements OneFS 9.2.0.0 includes enhancements to storage efficiency, quota statistics, and cluster statistics
to storage tracking to improve the accuracy of data reduction and storage efficiency reporting.
efficiency
reporting
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.1.0.0
Full release notes available at: https://dl.dell.com/content/docu100463
Supportability: SNAP_RESERVE_FULL, notifies the user when snapshot reserve space is The reserve space
CELOG nearly full. The threshold at which this event is triggered is now configurable, threshold can be modified.
and the reserve space threshold can now be changed, or customized in the CLI
and WebUI.
Supportability OneFS provides in-product license activation.
Supportability Upgrading OneFS is simplified with a bundle that includes the upgrade version Provides a smoother,
and Upgrade of OneFS and all applicable, necessary roll-up patches. less disruptive upgrade
experience.
Upgrade Upgrade performance improvements. Improves efficiency and
reduces latency during
upgrade.
Backup and OneFS adds multi-stream capability to the NDMP Backup Restart Extension Improves backup
Recovery (BRE). efficiency.
Antivirus OneFS adds support for CAVA-based antivirus solutions. Enables supporting a
variety of CAVA Antivirus
vendors.
CloudPools CloudPools performance improvements. Improved performance,
network bandwidth
utilization, and cost
savings.
CloudPools ECS2 (Virtustream) is no longer a supported CloudPools target.
Security: STIG The default behavior to redirect incoming communication from HTTP port 80 to Prevents use of all
HTTPS port 8080 is disabled. unnecessary and/or
non-secure functions,
ports, protocols, and/or
services.
Hardware OneFS 9.1.0.0 and later includes InfiniBand back-end network support for Expands back-end
PowerScale F200 and F600 nodes. support for the F200 and
F600 nodes.
See the PowerScale OneFS support pages on www.dell.com/support for detailed user documentation.
OneFS 9.0.0.0
Full release notes available at: https://dl.dell.com/content/docu98858
Isilon is now Isilon OneFS is now PowerScale OneFS. The brand change is reflected in the WebUI. The CLI isi
PowerScale commands and OneFS API are not affected by this change, though output strings are updated with the
PowerScale brand.
Basic S3 protocol OneFS 9.0.0.0 adds basic S3 data access protocol support as a multi-protocol option.
support in OneFS
Support for up to OneFS 9.0.0.0 supports Leaf-Spine networks for up to 252 nodes.
252 nodes in a
OneFS cluster
PowerScale F200 The PowerScale F200 and F600 nodes are all-flash, 1U models that require a minimum cluster size of
and F600 nodes three nodes. Clusters can be expanded to a maximum of 252 nodes in single node increments.
OneFS 9.0.0.0 supports inline software data compression and data deduplication on the F200 and F600
nodes. Inline data deduplication and software-enabled inline compression are enabled by default on F200
and F600 nodes.
The F600 node uses direct-attached NVMe (Non-Volatile Memory Express) SSDs for fast data access,
with built-in parallelism.
Security On new installations of OneFS 9.0.0.0 and later, all protocols are disabled by default. You must enable
enhancement: any protocols that you plan to use. Use the following command to enable a protocol using the CLI, where
protocols are <protocol> is one of smb, nfs, hdfs, ftp, http, https, or s3.
disabled by
default isi services <protocol> enable
Enhancements to OneFS 9.0.0.0 enables configuring and managing the HealthCheck tool using the OneFS 9.0.0.0 Web
the HealthCheck Administration UI. Use the WebUI to configure HealthCheck runs, manage notifications, review reports,
tool and more.
Support for When OneFS CloudIQ is enabled, telemetry data is sent to Dell Technologies using Secure Remote
enabling and Services (SRS). Data includes system logs and system configuration data. The data is used to build alerts
disabling CloudIQ and performance metrics to improve product diagnostics and issue resolution. CloudIQ is enabled by
on a per-cluster default. If you do not want to send telemetry data to Dell Technologies, you can disable CloudIQ using the
basis OneFS 9.0.0.0 Web Administration UI.
Support for OneFS 9.0.0.0 provides a Web Administration UI for managing pools of compatible node types. You can
managing node change only those pools that do not have hard restrictions.
pools and
compatible node
types in the Web
UI
OneFS Ranger The OneFS Ranger plugin is enhanced to support NANON for Hadoop.
plugin support
for NANON
New events and There are new OneFS events and alerts for 9.0.0.0. For details, see the PowerScale OneFS 9.0.0.0 Event
alerts Reference Guide.
OneFS 9.5.0.0
Full release notes available at: https://dl.dell.com/content/manual58071479-powerscale-onefs-9-5-0-0-release-notes.pdf
Known Issues
This section describes issues that are known to affect this release of OneFS.
For information about issues known to affect other OneFS releases, log in to the PowerScale OneFS Product Page and view the
available OneFS Release Notes.
Datamover/SmartSync
PSCALE-133734 The Datamover to Cloud services copy job can pause with the message:
Workaround: Restart the Datamover service and resume the copy job.
PSCALE-132042 The Datamover service can get stuck in the running state when attempting to copy an object to a Cloud
service with a path name longer than 702 characters. Contact Dell Support to terminate the Datamover
daemon and restart it.
PSCALE-129725 This known issue has two parts:
1. If you cancel a job within a minute of that job entering the pause state, Datamover may not honor the
job cancel request. The issue is the timing of the job cache refresh on the Datamover node.
Workaround: If the job is not canceled within a couple of minutes after sending the job cancel request,
send another job cancel request.
2. If the remote account for a copy job is not reachable and you send a request to pause the job, the
pause request will not be honored while the account is not reachable.
Workaround: Resolve the account connectivity issue and re-send the pause request
File System
PSCALE-154245 If the cluster has custom FTP or FTPS settings for isi_gather_info, and if the cluster loses quorum, or if
the /ifs partition becomes read-only, log uploads may fail.
Workaround:
Specify the --noconfig option when initiating the log gather. Example:
isi_gather_info –-noconfig
Or provide your unique FTP/FTPS credentials with the --noconfig option. Example:
isi_gather_info –-ftp –ftp-user <username> --ftp-host <ftp-host> --ftp-path
<path> --noconfig
PSCALE-153229 When a file policy name exceeds 199 bytes, the job engine enters an infinite core dump loop.
Workaround: Avoid creating file policy names that exceed 199 bytes.
PSCALE-128902 Changes in cluster configuration such as adding or removing nodes or disk pools becoming full, can result
in the creation of many small shadow stores. Most of these shadow stores are empty and will be removed
by the ShadowStoreDelete job. In some cases, the result can be a large number of shadow stores
added to a set of stale SINs and, rarer, a node panic.
Workaround: Run the following command daily:
This will avoid node panics, though the cluster will continue creating lots of small shadow stores.
Networking
PSCALE-162900 In rare circumstances, SmartConnect's internal caches can get out of sync with the FlexNet configuration
and have differing views of what interfaces and IPs belong in each network pool. The easiest way to view
this is by running the `isi network pool status <pool id> --show-all` command and noticing discrepancies
with the network pool configuration. The work around is restart SmartConnect on any affected nodes.
PSCALE-161675 On clusters that contain a large number of dynamic network pools, it is possible for a failover event
(such as a node reboot) to trigger more internal system events than SmartConnect can keep up with.
This can lead to high CPU usage in isi_smartconnect_d and isi_flexnet_d. It could also cause a crash
of isi_smartconnect_d. Outside of reducing the number of dynamic network pools, there is currently no
workaround.
Protocols
PSCALE-161316 When concurrent session limits feature is enabled, there is a potential for temporary user lockout if
cluster time is changed or the JWT signing keys are rotated. User access resumes when a previous
session expires.
PSCALE-160907 With API authentication mode configured to require SSO, user's without
ISI_PRIV_LOGIN_PAPI_BYPASS_MFA cannot authenticate to the WebUI using username and password
login form. Instead sign-in with SSO option.
PSCALE-160679 If STIG hardening profile is disabled, all NFS exports will have their security flavors reset from Kerberos
5 Privacy (krb5p) to the system default of Unix. If Linux clients are configured to use krb5p, they will
experience DU until they are configured to either use Unix (auth_sys) authentication, or the exports are
all re-configured to use krb5p.
Security
PSCALE-162981 If you applied the STIG hardening profile and disabled the STIG hardening profile, SSO MFA remains
enabled. This setting may prevent accounts without the ISI_PRIV_LOGIN_PAPI_BYPASS_MFA from
accessing the WebUI or PAPI.
Workaround:
To disable this setting, log on to the CLI as root user and run the following command:
OneFS 9.4.0.0
Full release notes available at: https://dl.dell.com/content/manual62544979-powerscale-onefs-9-4-0-0-release-notes.pdf
Known Issues
This section describes issues that are known to affect this release of OneFS.
For information about issues known to affect other OneFS releases, log in to the PowerScale OneFS Product Page and view the
available OneFS Release Notes.
Datamover/SmartSync
PSCALE-133734 The Datamover to Cloud services copy job can pause with the message:
Workaround: Restart the Datamover service and resume the copy job.
PSCALE-132042 The Datamover service can get stuck in the running state when attempting to copy an object to a Cloud
service with a path name longer than 702 characters. Contact Dell Support to terminate the Datamover
daemon and restart it.
PSCALE-129725 This known issue has two parts:
1. If you cancel a job within a minute of that job entering the pause state, Datamover may not honor the
job cancel request. The issue is the timing of the job cache refresh on the Datamover node.
Workaround: If the job is not canceled within a couple of minutes after sending the job cancel request,
send another job cancel request.
2. If the remote account for a copy job is not reachable and you send a request to pause the job, the
pause request will not be honored while the account is not reachable.
Workaround: Resolve the account connectivity issue and re-send the pause request
File system
PSCALE-128902 Changes in cluster configuration such as adding or removing nodes or disk pools becoming full, can result
in the creation of many small shadow stores. Most of these shadow stores are empty and will be removed
by the ShadowStoreDelete job. In some cases, the result can be a large number of shadow stores
added to a set of stale SINs and, rarer, a node panic.
Workaround: Run the following command daily:
This will avoid node panics, though the cluster will continue creating lots of small shadow stores.
Networking
PSCALE-132973, If you have a OneFS cluster with over 100 nodes, and are using redundant back-end Int-B Infiniband
PSCALE-117317, switches, and you encounter an issue that causes your back-end switches to reboot, nodes can split from
PSCALE-119564 the OneFS cluster. If this issue happens across a large number of nodes on the same cluster, it might
result in data unavailability on that cluster for a few seconds at a time.
Workaround: OneFS should retry adding the nodes to the cluster automatically. Once the nodes are
added to the cluster, the issue will be resolved.
SyncIQ
PSCALE-117552 SyncIQ is unable to rename a target directory if its parent directory has a "default-directory" quota set. If
there is a target subdirectory quota, SyncIQ provides a "not supported" message.
OneFS 9.3.0.0
Full release notes available at: https://dl.dell.com/content/manual64485109
Known Issues
This section describes issues that are known to affect this release of OneFS.
For information about issues known to affect other OneFS releases, log in to the PowerScale OneFS Product Page and view the
available OneFS Release Notes.
Hardware
PSCALE-116185 It is possible for FIPS SED drives to become unavailable after repeated suspend/re-add operations.
PSCALE-117511 An NVDIMM device may encounter a critical error, possibly resulting in system performance degradation.
The operating system and/or applications may fail as a result. There may be a message similar to the
following:
Networking
PSCALE-113065 Nodes A300, A3000, H700, and H7000 ship with BMC LAN access disabled. Because of a firmware
issue, the following commands to enable access may not work, and you may not be able to change
allocation-type from static to dhcp or from dhcp to static:
Protocols
Placeholder Placeholder.
Security
PSCALE-116997 Reverting security hardening can fail with the message:
Supportability
PSCALE-119865 In rare circumstances, running isi stats heat on a very large cluster with tens of thousands of files
accessed can time out.
Workaround: Run isi stats heat –numeric to get statistics on global inodes rather than file
names.
PSCALE-60739 In rare circumstances, there can be a performance issue when there are more than a thousand snapshot
queries using SNMP.
Upgrades
PSCALE-119827 When upgrading a hardened cluster to OneFS 9.3.0.0, the upgrade might fail if a previous upgrade
attempt was rolled back.
Workaround: If upgrading from OneFS 9.2.x, install the latest patch before upgrading to OneFS 9.3.0.0.
If upgrading from OneFS 9.0.x and 9.1.x, and you encounter an upgrade failure, run the isi hardening
apply --profile=STIG command, then run the upgrade to OneFS 9.3.0.0 again. If upgrading from
OneFS 8.2.2 and earlier, it is recommended to upgrade to OneFS 9.0.0.0 first, before upgrading to OneFS
9.3.0.0.
OneFS 9.2.1.0
Full release notes available at: https://dl.dell.com/content/docu104030
Known Issues
This section describes issues that are known to affect this release of OneFS.
For information about issues known to affect other OneFS releases, log in to PowerScale Technical Support and view the
available OneFS Release Notes
Antivirus
PSCALE-21576 If you use isi antivirus cava servers modify --server-name=<name> --server-
uri=<uri> to configure multiple CAVA server URIs and the interval between configurations is short,
the CAVA URI may not be configured correctly.
Workaround: Restart the lwavscan daemon.
PSCALE-23191 CAVA jobs always start on the first node in the cluster. If Node 1 fails or splits from the cluster, no
scheduled CAVA jobs can start until Node 1 rejoins the cluster. Running jobs are not affected.
Workaround: The issue resolves itself after Node 1 rejoins the cluster.
File system
PSCALE-48358 The following warning message can appear when the backup of multi TB files or data sets fails:
PSCALE-23063 The following warning message can appear repeatedly on the console:
Console spam - WARNING: Intel (R) Network Connections are quality tested
using Intel (R) Ethernet Optics.
NFS
PSCALE-57485 In rare circumstances, NFSoRDMA may not work after a node reboot.
Workaround: Restart the NFS service.
PSCALE-37664 If the sunrpc.rdma_slot_table_entries parameter is set to 128 or higher, the cluster might experience
write performance slowness.
Workaround: Do not set the sunrpc.rdma_slot_table_entries parameter to 128 or higher.
Security
PSCALE-60478 Upgrade from OneFS 8.1.2 to 9.2.0.0 or 9.2.1.0 does not preserve the HTTP basic auth setting.
Workaround: Before you upgrade, install the latest 8.1.2 RUP.
Upgrades
PSCALE-25579 If you are upgrading to OneFS 9.2.0.0 or later, CELOG maintenance mode is no longer supported. If you
are upgrading to OneFS 9.2.0.0 or later, suppressed mode might cause the upgrade to fail.
Workaround: Take your cluster out of maintenance mode before upgrading to OneFS 9.2.0.0 or later. Do
not put any nodes into suppressed mode until the upgrade has been committed.
PSCALE-22265 When upgrading OneFS, if your cluster is displaying critical hardware events, using the isi event bulk
--ignore or isi event bulk --resolve commands might not ignore or resolve the events and cause the
upgrade to fail.
No workaround.
Known issues
This section describes issues that are known to affect this release of OneFS.
For information about issues known to affect other OneFS releases, log in to Dell Technologies Support and view the available
OneFS Release Notes.
Antivirus
PSCALE-21576 If you use isi antivirus cava servers modify --server-name=<name> --server-
uri=<uri> to configure multiple CAVA server URIs and the interval between configurations is short,
the CAVA URI may not be configured correctly.
Workaround: Restart the lwavscan daemon.
PSCALE-23191 CAVA jobs always start on the first node in the cluster. If Node 1 fails or splits from the cluster, no
scheduled CAVA jobs can start until Node 1 rejoins the cluster. Running jobs are not affected.
Workaround: The issue resolves itself after Node 1 rejoins the cluster.
PSCALE-56419 The CAVA anti-virus server encounters a connectivity issue in an IPv6 environment.
Workaround: Use CAVA only in an IPv4 environment.
File system
PSCALE-48358 The following warning message can appear when the backup of multi TB files or data sets fails:
Networking
PSCALE-55830 If a node fails to mount the /ifs directory, the following error message can appear:
PSCALE-23063 The following warning message can appear repeatedly on the console:
Console spam - WARNING: Intel (R) Network Connections are quality tested
using Intel (R) Ethernet Optics.
NFS
PSCALE-57485 In rare circumstances, NFSoRDMA may not work after a node reboot.
Workaround: Restart the NFS service.
PSCALE-37664 If the sunrpc.rdma_slot_table_entries parameter is set to 128 or higher, the cluster might experience
write performance slowness.
Workaround: Do not set the sunrpc.rdma_slot_table_entries parameter to 128 or higher.
Upgrades
PSCALE-59530 When upgrading the cluster to OneFS 9.2.0.0, some eventgroups that occurred before or during the
upgrade will show up as having an "Unknown" cause. Some common events include reboot, synciq and
external_network, but can also include avscan, filesystem, hwmon, overheating, powersupply, physmem,
reboot-fail, storage_transport, synciq, windwos_auth, windows_idmap, and windows_networking. If new
events occur after these "Unknown" eventgroups are resolved, they will show up with the correct
message.
Workaround: The situation resolves itself after upgrade after the "Unknown" events are set to
"Resolved".
PSCALE-25579 If you are upgrading to OneFS 9.2.0.0 or later, CELOG maintenance mode is no longer supported. If you
are upgrading to OneFS 9.2.0.0 or later, suppressed mode might cause the upgrade to fail.
Workaround: Take your cluster out of maintenance mode before upgrading to OneFS 9.2.0.0 or later. Do
not put any nodes into suppressed mode until the upgrade has been committed.
PSCALE-22265 When upgrading OneFS, if your cluster is displaying critical hardware events, using the isi event bulk
--ignore or isi event bulk --resolve commands might not ignore or resolve the events and cause the
upgrade to fail.
No workaround.
OneFS 9.1.0.0
Full release notes available at: https://dl.dell.com/content/docu100463
PSCALE-20905 Protocols: Antivirus SMB may become inaccessible if you enable the To avoid this issue,
CAVA service by running the command isi use isi antivirus
service cava enable. cava settings
modify --service-
To resolve this issue, run the following commands:
enabled=yes to enable
1. isi antivirus cava settings modify the CAVA service.
--service-enabled=no
2. isi antivirus cava settings modify
--revert-ip-pool
3. isi zone zones delete AvVendor
4. isi antivirus cava settings modify
--ip-pool=groupnet0.subnet0.pool1
5. isi antivirus cava settings modify
--service-enabled=yes
PSCALE-18859 Protocols: Antivirus It is possible for the CAVA daemon (lwavscan) to No action is required
crash while shutting down if there are outstanding when this occurs.
scan requests in its request queue. This generates a
core file, but does not affect the subsequent restart
of the daemon or normal operation of antivirus
scanning afterwards.
PSCALE-18830 FreeBSD: Kernel When rebooting a node, it is possible for NFS to If you encounter this
survive quorum loss, which might prevent a new issue, you must manually
instance of NFS from starting up on that node. If this stop NFS on that node.
issue occurs, var/log/messages will display the Stopping NFS will allow a
following message: new instance of NFS to
start.
service nfs already open by pid
.
PSCALE-18691 Protocols: NFS If an NFS client tries to connect to a cluster using the Users should avoid
Access Zone called AvVendor, the NFS daemon will connecting to the cluster
fail. with the AvVendor
Access Zone. When
CAVA antivirus is enabled,
the Access Zone called
AvVendor is created
PSCALE-15733 Upgrade and During an upgrade from OneFS 9.0.0.0 to OneFS If you encounter this
Installation 9.1.0.0, a node might enter a state that keeps the issue, you must manually
node from rebooting. If this issue occurs, it might reboot any affected
cause the upgrade to stall. nodes after other nodes
in the cluster have
finished upgrading.
PSCALE-15298 Protocols: Antivirus The isi antivirus cava status Review your antivirus
-v command and isi statistics software's logs.
query current --format json --keys
node.antivirus.statscommand output does
not count the number of CAVA jobs under Scan Fails
in the scan report. The count is provided only for
on-access scan.
PSCALE-14506 Protocols: Antivirus If you create a CAVA job with one non-existing path
and multiple valid paths, the job never starts. You can
find the following message in the lwavscand.log.
InvalidObjectState (403)
OneFS 9.0.0.0
Full release notes available at: https://dl.dell.com/content/docu98858
File system
263645 In rare circumstances, when inline deduplication is enabled on large clusters, the performance of nodes
that have a mirror of a certain shadow store reference count block may become limited.
Workaround: Inline dedupe can be enabled on large clusters but you should disable the
efs.sfm.inline_dedupe.do_read sysctl. Use the following command to disable the sysctl:
Hardware
267404 After you smartfail a drive, remove the drive from the system, and then reinsert the drive, false power
supply events may occur. This is due to a known firmware issue which will be addressed in a June, 2020
update.
Networking
272321 A rare race condition in the InfiniBand driver or a faulty InfiniBand card can cause a node to fail with a
panic message:
Stack: --------------------------------------------------
kernel:trap_fatal+0xa2
kernel:trap_pfault+0x2fd
kernel:trap+0x61d
kernel:ib_create_ah+0x9
kernel:cm_alloc_msg+0x33
kernel:ib_send_cm_dreq+0xbe
kernel:rdma_disconnect+0x149
kernel:sdp_output_reset+0xa4
kernel:sdp_shutdown_task+0x96
kernel:taskqueue_run_locked+0xe5
kernel:taskqueue_thread_loop+0xa7
kernel:fork_exit+0x74
Workaround: Rebooting the node will resolve the issue if the issue was caused by the rare race
condition. If rebooting the node does not resolve the issue, a faulty InfiniBand card may be the cause.
Contact Dell Technologies Supportto get help with replacing the card on the node.
ps -uax | grep lw
If lwsmd is the only service listed and has not logged any messages to /var/log/lwsmd.log since
it stopped services, it is possible that your node is in this state. To resolve the situation, kill the lwsmd
process.
272919 If you use the S3 protocol with SmartConnect, and a zone's base domain is not set using the isi s3
zone settings modify command, the client may receive signature mismatch and other errors.
Workaround: To avoid this issue, ensure to set the base domain using the isi s3 zone settings
modify command.
272713 If ESRS with telemetry is enabled and NFS is not enabled, the isi_esrs_d logs will show an error when
gathering telemetry data for vfs.nlm and nfs/log-level. You can ignore these errors.
272430 An Active Directory username specified during the S3 secret key generation does not give a user access
to S3 unless it is specified in the canonical format of DOMAIN\user, where DOMAIN is uppercase, and
user is lowercase.
Additionally, an Active Directory username specified in the ACL of a bucket does not give that user access
unless it is specified in the same canonical format.
272287 On certain occasions, the S3 service associated with the cluster's health conditions may not be ready
to accept new connections and requests for a brief period. As a result, you may experience data
unavailability (DU).
272018 If a multipart upload is performed on a directory for which the bucket owner does not have access
permission, the bucket owner cannot stop the upload or list parts.
271215 When you try to list multipart uploads using Delimiter='/', groups are not broken into common prefixes.
271214 When listing multipart uploads, the Delimiter field is silently ignored, and the slash character is always
used as a delimiter.
271084 The S3 protocol does not include packet header sizes in bandwidth usage statistics.
271027 Etag changes on the SyncIQ target cluster.
268952 If Bucket Directory Create Mode for a zone is configured without the execute permission, clients cannot
access the bucket.
Workaround: To avoid this issue, ensure to add the execute permission when changing the Bucket
Directory Create Mode using the isi s3 settings zone command.
267883 The S3 protocol does not support a file name which is not represented as a valid UTF-8 string. As a result,
the following error might appear:
267549 Administrators can delete S3 buckets that are not empty. This behavior differs from the S3 protocol,
which does not allow non-empty bucket deletion. OneFS handles bucket deletion the same way that SMB
share and NFS export deletion are handled. When you delete an S3 bucket through OneFS, the bucket
path is deleted but the directory it referenced still exists. If you create a new bucket with the same path
as the bucket that was deleted, the directory that the previous bucket referenced will be accessible again
through the new bucket.
266840 If S3 bucket operations like Put-Bucket and Delete-Bucket go through the protocol layer, they are not
audited.
InvalidObjectState (403)
264415 The S3 protocol does not validate the Content-Md5 request header. To avoid calculating Md5, a
synthetic etag is created when Content-Md5 is not supplied in the request.
263655 While creating or modifying a bucket from the OneFS web administration interface, you can add the same
access control list (ACL) multiple times to the bucket.
263571 The S3 protocol does not support aws-chunked (streaming upload) content-encoding.
262949 If you try to remove a bucket through PAPI when the S3 service is disabled, the bucket gets deleted.
But, the temporary directory (.isi_s3_xxx) under the bucket root directory and part directories
(.isi_s3_parts_xxx) under the target directory does not get deleted. You must remove these
directories manually.
SmartPools
271386 When adding A100 nodes to a cluster or when forcing reprovisioning of A100 nodes, it is possible that a
CELOG event similar to the following appears:
When a node is not provisioned, that means that it is not associated with a node pool. You cannot write to
a node that is not provisioned.
Workaround: Attempt to move the unprovisioned A100 nodes into the same node pool by using
compatibilities. See the OneFS Web Administration Guide or OneFS CLI Administration Guide for details.
SMB
267941 If a client connects to an SMB3-CA share using a local account, automatic migration during a planned
maintenance event may not happen correctly. This can increase the length of an I/O pause, although it
will not lead to a data unavailable condition.
Workaround: To avoid this issue, ensure that the cluster has been joined to the same Active Directory
domains as the SMB3-CA clients.
HookDataMigrationUpgrade|Failed:7
268043 If you perform a rollback of a cluster upgrade and, while the cluster is archiving, issue an isi_for_array
hostname command, it might cause the rollback process to stall.
Workaround: If you encounter this issue, extend the isi_for_array timeout by running the following
command: isi_for_array --rpc-timeout 180
263910 If you perform a parallel upgrade from OneFS 8.2.2 to OneFS 9.0.0.0, you might encounter an issue where
all nodes on a network appear to be unavailable. If this issue occurs, the upgrade fails.
Workaround: Use the --skip-optional option when you restart the parallel upgrade.
261596 If you perform a rolling upgrade of your OneFS cluster, and you have dynamic IPs configured,
SmartConnect might not allocate them properly after the upgrade.
Workaround: Manually move the dynamic IPs after the upgrade.
OneFS Web UI
272790 The WebUI does not search hidden directories. For example, if you place directories or files in a hidden
directory such as /ifs/.secretdir and then browse for them using the WebUI, you will not be able to
find them.
Workaround: Manually add the proper path, or do not place directories or files in hidden directories.
271291 In some cases, the WebUI does not display complete IPv6 addresses.
Workaround: To view the complete IPv6 addresses for the internal network, go to the Network
Configuration > Internal tab. To view the complete IPv6 addresses for the external network, go to
the Network Configuration > External tab and click View Pool Details.