Hitachi Content Software for File v4.0.1.36 Release Notes
About this document
This document (RN-HCSF003-03, October 2022) provides late-breaking information about Hitachi Content Software for File v4.0.1.36. It includes information that was not available at the time the technical documentation for this product was published, as well as a list of known problems and solutions.
Intended audience
This document is intended for customers and Hitachi Vantara partners who license and use Hitachi Content Software for File.
Accessing product documentation
Product user documentation is available on the Hitachi Vantara Support Website: https://knowledge.hitachivantara.com/Documents. Check this site for the most current documentation, including important updates that may have been made after the release of the product.
Accessing product downloads
Product software, drivers, and firmware downloads are available on the Hitachi Vantara Support Website: https://support.hitachivantara.com/.
Log in and select Product Downloads to access the most current downloads, including important updates that may have been made after the release of the product.
About this release
This document describes the differences between the following versions.
Product | New Version | Previous Version |
Hitachi Content Software for File | 4.01.36 | 3.12.3.12 |
Release highlights
Highlights in v4.0.x
Highlights in v3.14.x
Highlights in v3.13.x
Breaking Changes / Deprecation in v4.0.x
Breaking Changes / Deprecation in v3.14.x
Breaking Changes / Deprecation in v3.13.x
New features
The following table lists and describes the new features in HCSF v3.13.x, v3.14.x, and v4.0.x.
Item | Description |
WEKAPP-190497 | Added dark mode to the GUI Added dark mode to the GUI so the user can switch between the light and dark modes. |
WEKAPP-194475 | CLI to get information about uploaded snapshots from a bucket Introducing a new CLI command weka fs tier s3 snapshot list, that given an object-store bucket name and a snapshot locator provides some basic information about the uploaded snapshot of that locator. |
WEKAPP-224245 | Client Quality of Service (QoS) using mount options Introduced new mount options (qos_max_throughput_mbps and qos_preferred_throughput_mbps) to control the max and preferred throughput rate of a client host. |
WEKAPP-232466 | Snapshot upload task abort Snapshot upload tasks can now be aborted via CLI (weka cluster tasks abort) and via REST API. |
WEKAPP-232467 | Snapshot upload pause/resume support Snapshot upload tasks can now be paused/resumed via CLI (weka cluster tasks pause/resume) and via REST API. |
WEKAPP-232797 | Snapshot upload size Added 2 new fields to the snapshots list output, that show the amount of data that needs to be uploaded to the object store in order to upload each snapshot (one for the local object-store bucket and the other for the remote object-store bucket). |
New functionalities and improvements
The following tables list the new functionalities and improvements in HCSF v3.13.x, v3.14.x, and v4.0.x.
WEKAPP-63980 | Filesystem removal completion event Added an event to indicate when filesystem removal completes. |
WEKAPP-171231 | Additional information on drive failure |
Added additional drive failure information to both the weka cluster drive CLI command output and DriveDown alert. | |
WEKAPP-177502 | Snapshot upload/download tasks are now listed under cluster tasks Snapshot upload/download tasks are listed under the weka cluster tasks CLI table. Note: pause/resume is not yet supported for these tasks. |
WEKAPP-200365 | Entities UID listing in the CLI using verbose flag CLI commands that list entities (such as weka cluster host, weka cluster nodes, etc.) now have a column of unique ID (that can be used to interact with the product's REST API). The column is displayed only with the -v|--verbose CLI flag. |
WEKAPP-201141 | df utility integration with quotas When a hard quota is set on a directory, running the df utility will consider the hard quota as the total capacity of the directory and provide the use% relative to the quota. This can help users understand their usage and how close they are to the hard quota. Note: The df utility behavior with quotas is currently global to the system. To change the global behavior, contact customer support. |
WEKAPP-207557 | Additional information about local system processes Provide additional information on the local system processes running on the host (using weka local status CLI command). |
WEKAPP-211218 | POST cluster REST API Added the REST API route to form a new cluster. |
WEKAPP-212694 | Stateless client mount syntax Added an option to mount wekafs filesystem with similar syntax to NFS syntax (use of : between the server and filesystem, in addition to the existing syntax using /). Supported mount syntax: mount -t wekafs <backend>/<filesystem> <mount-point> mount -t wekafs <backend>:/<filesystem> <mount-point> |
WEKAPP-215924 | System Status: REDISTRIBUTING The system will now appear as REDISTRIBUTING (instead of REBUILDING) when the system is redistributing data while being fully protected (for example, due to graceful introduction or removal of drives). |
WEKAPP-217086 | Directory quotas REST API for new directories Added REST APIs for directory quotas. the PUT and DELETE APIs will only work on empty directories. To apply directory quota on an existing directory with data, the CLI should be used. |
WEKAPP-219616 | Allow an immediate removal of a (down) host Introduced --no-unimprint option to weka cluster host remove that allows removing a host from the cluster configuration without connecting to it and reverting it to STEM mode. This is useful for removing a host that is down immediately without waiting for the cluster to finish unimprinting it. |
WEKAPP-223503 | Warning when stopping local system processes Cannot stop local system processes (using weka local stop CLI command) if a filesystem is mounted on that host. Use --force to bypass. |
WEKAPP-224356 | More informative hanging IO alerts and events Hanging IOs alerts and events now report: The inode the IO is associated with The state the IO is hung in The pid of front-end hung IOs |
WEKAPP-225084 | Added support to enable/disable the login banner Use the Login Banner menu in the GUI or weka security login- banner enable/disable CLI command. |
WEKAPP-227148 | Limit concurrent snapshot uploads Limit the number of concurrent snapshot upload tasks to one per object- store. There will be only one snapshot (per object-store) using the resources and it will get uploaded sooner, rather than have up to 16 snapshots (of different filesystems) competing on the same resources, delaying the upload completion for all. |
WEKAPP-230910 | Setting encrypted filesystem only if a KMS is configured Disable the option to set an encrypted filesystem in the GUI if a KMS is not configured. |
WEKAPP-231427 | Alert about SSD capacity usage The alert about SSD capacity usage was modified to only alert when the SSD overall capacity is overflowing and utilizing internal capacity buffers. |
WEKAPP-231786 | CLI output changes to boolean column Changed text of the column Muted in weka alerts CLI command to show Muted or blank, as opposed to True/False. Changed text of the column Dedicated in weka cluster host CLI command to show Dedicated or blank, as opposed to True/False. Changed text of Upload Tags Enabled columns in OBS CLIs to show Enabled/Disabled as opposed to True/False. |
WEKAPP-240718 | Support installing kernel-ml on RHEL/CentOS Added support for installing mainline kernel packages (kernel-ml) on RHEL/CentOS. |
WEKAPP-244356 | Increased number of the local users The total number of the local users is increased to 1152 (previously, it was 512). |
WEKAPP-245464 | Added support for LDAP over SSL Added support for LDAP over SSL (LDAPS). |
WEKAPP-249124 | Object-store operations list Enhancement: The command weka fs tier ops runs across all object store buckets when the object store bucket name is not specified. The command output includes the last HTTP error code an operation encountered. |
WEKAPP-249499 | Added support for Mellanox OFED 5.6-1.0.3.3 driver |
WEKAPP-254139 | weka cluster CLI commands use the singular form The weka cluster CLI commands use both singular and plural forms: Instead of weka cluster nodes, use weka cluster node Instead of weka cluster tasks, use weka cluster task Instead of weka cluster buckets, use weka cluster bucket Apart from weka cluster host, the weka cluster hosts CLI command is also supported. The plural form of these commands still works as an alias for backward compatibility |
Fixed issues
The following tables list the fixed issues in HCSF v3.13.x, v3.14.x, and v4.0.x.
Clustering
Item | Description |
WEKAPP-224233 | Resources restore CLI command The weka local resources restore command does not correctly restore the resources. |
WEKAPP-225760 | Adjusted mount behavior for non-supported mount option by networking device When mounting, prevent and block setting dedicated_mode=NONE when RX interrupt is not supported by the network device Device with DPDK networking. |
WEKAPP-227771 | Un-mounting during upgrade can cause the upgrade to fail Un-mounting a mount-point from a host that is part of the cluster during an upgrade can prevent the host from joining back to the cluster and might cause the upgrade to fail. Impact: Upgrade failure Configuration: Converged environments; Legacy clients |
WEKAPP-228463 | Cannot start-io after changing IPs on all hosts The system might fail to start-io after changing IPs for all of the hosts. |
WEKAPP-229859 | Incorrectly adding new network devices in cpu-optimized mode (UDP) When manually adding network devices after the cluster is up and running, they were added in UDP mode instead of DPDK. Impact: Reduced performance |
WEKAPP-230789 | Thin-provisioned filesystems in multiple organizations Miscalculation of free space in non-root organizations without an organization quota and thin-provisioned filesystems might lead to failures in creating new (thick provisioned) filesystems in that organization. Configuration: Multiple organizations and thin-provisioned filesystems |
WEKAPP-234726 | Clients connectivity not recovering after heavy load In some circumstances, under heavy load, some clients might experience permanent degraded connectivity instead of a transient degradation. |
WEKAPP-244236 | Add a backend host failure When adding a backend host using the weka local setup command, the backend host may fail to join the cluster during a race condition. |
Clustering / Agent
Item | Description |
WEKAPP-205068 | Reduced excessive logging to Syslog Reduced the amount of log messages that log the output of the NFS server get_command_output command to Syslog. Configuration: Hosts configured with the new NFS stack |
WEKAPP-215939 | Reduce messages generated to Syslog Removed redundant DEBUG: Answering if container samba is running messages to Syslog. |
WEKAPP-215956 | Mount failure due to agent failure during startup The agent process might crash on startup in case of an ungraceful shutdown. Impact: Mount failure |
WEKAPP-217244 | Mount option dedicated_mode using lowercase value resulting in failure Mount option value for dedicated_mode only accepts uppercase - either "NONE" or "FULL" instead of other case insensitive variations, as it should. Impact: Mount failure |
WEKAPP-235449 | Updated Syslog level for mount and umount operations The Syslog log level of the mount and umount commands is now DEBUG, improving diagnostic capabilities. |
WEKAPP-237963 | Resolved issue where system agent may restart System agent might restart under concurrent API calls load. |
WEKAPP-242300 | API Output log rotation API output logs (api-stdout.log) were not rotated, leading to logs directory becoming completely full. |
WEKAPP-242464 | Monitor the capacity of the logs Added alerts and events indicating when the logs have low disk space. |
Clustering / Failure Domains
Item | Description |
WEKAPP-261888 | Cannot add a drive to a host that's on a different failure domain The system does not prompt a message and does not prevent adding a drive to a host if the drive was previously configured with a different failure domain. The solution includes adding a relevant message and preventing this action. |
Control / API
Item | Description |
WEKAPP-205945 | REST API /security/tokensExpiry route rename Change route name of /security/tokensExpiry (deprecated) to /security/defaultTokensExpiry, to avoid confusion. |
Control / Authentication
Item | Description |
WEKAPP-216114 | Organization creation password complexity enforcement When creating a new organization, along with its org-admin, the system's password complexity policy is not enforced. |
WEKAPP-238869 | Enabled usage of a user-obtained token during sudo operations. HCSF does not allow an interactive login during mount, and the user must log in before the mount command when working with a filesystem with authentication-required. The weka user login CLI command saves the auth-token to the user's home directory, whereas sudo mount will look for the auth- token in the root user home directory. The result will be that users will perform a login, try to mount, and will be told to log in even though they're logged in. Impact: Mount failure |
WEKAPP-250579 | LDAP/AD usernames with special characters Setting up an AD/LDAP server for system management fails if it includes usernames with special characters. |
Control / Config
Item | Description |
Failures in the cluster leader when in the stopped IO state (as during an upgrade) might result in drives not identified for phasing out, IO not being able to start and causes the upgrade to fail. | |
WEKAPP-224981 | Unavailability issue due to large config change sets Under a rare statistical race condition in the system management process, due to a very large configuration delta being created, exceeding an internal buffer that was incorrectly defined being too small. This causes processes to continuously crash and for system unavailability. Impact: API and Data unavailability Configuration: More than 800 hosts (clients and backends) |
WEKAPP-225593 | Cluster initialization failure During cluster initialization, if a host that has drives fails to apply its resources correctly and a consecutive initial start-io is invoked, the cluster will never be able to start serving IOs. |
WEKAPP-236040 | Stateless client deactivation When the weka cluster host deactivate CLI command is invoked on a stateless client, the client rejoins the cluster after a few seconds as active. |
WEKAPP-246401 | Process restarts on role changes If the internal system process changes its role from Compute to FrontEnd, it can cause other processes to crash when they join the system bucket council that the process was once a member of. |
WEKAPP-253026 | CLI command weka status is unavailable If the cluster contains unresponsive buckets, the CLI command weka status can be unavailable. |
Control / Diags
Item | Description | |
WEKAPP-215875 | Diagnostic uploading failure Diagnostic uploading fails, if the current host was not chosen to collect diagnostic information from. | |
When running the command 'weka diags upload --clients' without specifying the host running the upload using the '–host-id' option, the upload will fail on an assertion error | ||
Control / Diags
Item | Description |
WEKAPP-212031 | "ObjectStoreStatusChanged" event was split into "ObjectStoreStatusUp" and "ObjectStoreStatusDown" events "ObjectStoreStatusChanged" of severity INFO used to report both UP and DOWN status changes. Instead, the new event "ObjectStoreStatusUp" with severity INFO and the new event "ObjectStoreStatusDown" with severity MAJOR will be reported. |
Control / Licensing
Item | Description |
WEKAPP-221941 | License display error The weka cluster license CLI command sometimes shows parsing errors instead of displaying a properly formatted error |
Control / Security
Item | Description |
WEKAPP-228495 | Authenticated Mount When using a filesystem with auth-required (this is required when using multiple organizations for the non-root organization), if the mount command is run by the user without supplying the authentication token, the interactive mount command asking for user/password will create an authentication token and store that token in the root user directory without removing that token later. We changed the behavior to not allow interactive login via the mount command (which is also problematic when mounting via fstab/systemd). Users should obtain an authentication token beforehand (via weka user login) and provide it to the mount command (with the auth_token_path option). |
Control / Stats
Item | Description |
WEKAPP-168666 | Statistics output sorting The weka stats … --sort=stat,value CLI command sorts the output lexicographically instead of numerically. |
WEKAPP-192827 | Alert on low local disk space on the host Raise an alert and issue an event when hosts reach a low local disk space threshold. These warn what services might be impacted. When enough disk space is freed, the alert is cleared (and a corresponding clear event is issued). |
WEKAPP-230113 | Incorrect units in realtime CLI stats The weka stats realtime CLI command incorrectly displays MS (milliseconds) instead of "μS" (microseconds). |
Control / Traces
Item | Description |
WEKAPP-217203 | Upgrade failure due to local disk out-of-space Existing core files are not rotated correctly and might cause upgrade issues if exist and those files consume too much space. |
Devops / Provision
Item | Description |
WEKAPP-252758 | Periodic restarts when removing an SMB host Removing a host part of an SMB cluster causes periodic restarts of management processes. |
Feature / Mounting
Item | Description |
WEKAPP-217545 | Misleading error message when not specifying the network interfaces when mounting on a physical server Mounting on a physical server without the net option should not be possible and should yield an appropriate error message. Instead, the command issues a wrong error message. |
WEKAPP-242126 | Stateless client mount The client receives a list of IP addresses from the backend and always tries even recent invalid IP addresses, which can lead to long mount times and timeouts. As part of the fix, two new mount options are introduced to control different timeouts during the mount operation: -- connect_timeout_secs and --response_timeout_secs Impact: Stateless client mount can take a long time if there are incompatible IP addresses in the backend |
Filesystem
Item | Description |
WEKAPP-55041 | Manual Prefetch Can Skip Content When Snapshots Are Used Manual prefetch may skip content due to snapshot and prefetch only some of the content, as some of the read I/Os after the prefetch will still be directed to the object tier. |
WEKAPP-220517 | umask applied on default-acl umask is not ignored when applying default ACL (not adhering to the POSIX standard). |
WEKAPP-222922 | Snapshot upload during object-store detach from filesystem might lead to cluster unavailability Uploading a snapshot to a remote object-store from a downloaded filesystem that is undergoing detach from object-store might lead to system processes restarts and cluster unavailability. |
WEKAPP-224783 | Unavailability after snapshot deletion A race condition between RAID shrinkage and background snapshot deletion processes can cause an internal inconsistency that leads to system compute-processes restarting. This can lead to full cluster unavailability. |
Impact: Unavailability | |
WEKAPP-228311 | Remote snapshot upload during migration Remote snapshot upload during migration of a downloaded filesystem can fail to upload file content from the downloaded filesystem. This may cause a loss of data in the remote copy, which means you cannot recover from the remote backups. The detach operation should be completed before uploading snapshots to the remote object store. |
WEKAPP-251689 | Inaccurate capacity indications The system can indicate inaccurate capacities due to rounding when converting values from bytes to MB, GB, TB, and so on. |
Filesystem / Capacity
Item | Description |
WEKAPP-214710 | Quota Accounting A rare race condition might lead to users being able to consume more space than the quota allocation allows by conducting a false subtraction. |
WEKAPP-216771 | Exceeding directory quota limit Multiple writers to the same directory quota can sometimes fill the quota beyond the hard limit. |
WEKAPP-219781 | Directory quota set/unset failure Setting or unsetting a directory quota fails if the directory contains symbolic links. |
WEKAPP-220765 | Writing above the filesystem SSD limit Clients might be able to write some data above the filesystem SSD capacity limit. IF the entire SSD is occupied, this might come at the cost of the hot-spare capacity, reducing the ability to complete a rebuild and restore resiliency after a failure. |
WEKAPP-225118 | Filesystem total capacity corrected reporting When SSD capacity is full, the system fails to change the filesystem's total capacity. |
WEKAPP-231509 | Removing a directory quota Failure to remove a nested directory quota after it has canceled/failed during the setting process |
WEKAPP-240948 | Quota REST API returns a wrong inode_id Directory quotas REST API returns the internal inode ID instead of the inode context (user-facing ID). |
WEKAPP-252153 | Incorrect high drive capacity alert and event HighDrivesCapacity alert and UsedSSDCapacityCriticalOverflow event can be incorrectly issued temporarily when too much capacity is removed (failed drives/hosts). Replace the capacity to cease oversubscribed errors. |
WEKAPP-253057 | Hanging IOs after setting nested directory quotas Setting a nested quota while it is used may cause clients' IOs to hang when the quota becomes active. |
WEKAPP-258165 | Setting directory quotas Directory quota set/unset fails if the directory contains UNIX sockets. |
WEKAPP-268022 | IO operations hang when a directory quota reaches its exact value When the directory quota is filled to its exact value, the QuotaHardLimitReached event is not issued, and IO operations to this directory may hang. |
WEKAPP-269936 | Default quota REST API returns a wrong inode_id Default directory quota listing REST API returns the internal inode ID instead of the external one. |
Filesystem / Clusterwide Tasks
Item | Description |
WEKAPP-232512 | Resuming an object-store detach task while it is still pausing Resuming a cluster task (object-store detach) while it is still in PAUSING state (visible in weka cluster tasks CLI) might cause the task to resume in an incorrect place. This might lead part of the detach task to be skipped, leading to data unavailability. |
Filesystem / File Locks
Item | Description |
WEKAPP-238775 | IOs might hang longer than required In some cases, locks recovery might wait for longer than the locks grace time, which is noticeable by the clients' applications. |
Filesystem / File Logic
Item | Description |
WEKAPP-214538 | False drive failure events A single IO core serving both COMPUTE and Front End (FE) processes might cause false identification of drive failures, which can lead to false events about drive failures or TooManyFailures. |
WEKAPP-221291 | Stale file attributes cache In rare circumstances, stale file attributes might be returned from cache. |
Filesystem / Infrastructure
Item | Description |
WEKAPP-238022 | Memory leak on full capacity System process may restart when a directory quota or a filesystem is at full capacity due to a memory leak. |
Filesystem / OBS
Item | Description |
WEKAPP-60978 | OBS Prefetch fetches files more accurately When configuring prefetch-mib in weka fs tier s3 add, it now prefetches files better and increases the performance of files residing in the object- store. |
WEKAPP-135311 | Prefetching a Large File from OBS In rare instances, using weka fs tier fetch CLI command to prefetch a large file from object store to SSD may result in a node failure event (ASSERT failed: mbuf pool is empty (pool dumped to file). |
WEKAPP-216932 | Updating object-store bucket site parameter Remove support for directly updating the object-store bucket site, as it should always be the same as the object-store site it's attached to. To change the site remove and add a new object-store bucket instead, with the desired object-store. |
WEKAPP-217406 | Filesystem download from multiple object-store buckets When downloading a filesystem from an object-store, if the object-store is specified as the additional-obs-bucket, while the snapshot locator is only present in one object-store, the download task may be stuck. |
WEKAPP-221605 | Changes in local IP binding behavior for OBS connections For REMOTE OBS buckets, IP binding will no longer be used for connections. This is important for connections to REMOTE object stores that reside in an external network that is not accessible by connections bound to a local NIC's IP address and/or subnet. |
WEKAPP-222344 | Remote snapshot upload from a downloaded filesystem If a downloaded filesystem is detached from the local object-store (containing the snapshot) and attached to a remote object-store instead, uploading snapshots to the remote object-store might not contain all the snapshot data. |
WEKAPP-222658 | Changed the default object-stores names The default object-stores names have been renamed from "default" & "remote_default" to "default-local" & "default-remote" respectively. |
WEKAPP-228356 | Uploading small files to object-storage might under-utilize available bandwidth Uploading small files to object-storage might not utilize the entire available bandwidth to the object-store, due to overestimation of the RAM required for small file uploads. |
Filesystem / RAID
Item | Description |
WEKAPP-209705 | False Hot Tier Near Capacity alerts Hot Tier Near Capacity alert might be issued incorrectly when the SSD capacity is not full (as can be seen in the weka fs CLI command output), and contain an incorrect percentage value for the drive fill leve |
WEKAPP-217232 | Cluster unavailability after upgrade When upgrading to 3.9.x, 3.10.x, or 3.11.x versions that are not including this fix, under very rare circumstances, the cluster might reach an internally inconsistent state resulting in cluster unavailability. Impact: Data Unavailability |
WEKAPP-220085 | Cluster unavailability Under very rare circumstances involving both drive and compute processes failures, the cluster might become unavailable. |
WEKAPP-220807 | Under rare circumstances, overwrites might get corrupted Under rare race conditions when overwriting a file during space reclamation or releasing it to the object-store, the overwrite might get corrupted. Configuration: Tiered filesystems |
Filesystem / Snap2Obj
Item | Description |
WEKAPP-229880 | Snapshot upload from downloaded filesystem to remote object-store might hang Uploading a snapshot to a remote object-store bucket from a downloaded filesystem might hang if the filesystem has not begun detaching from the local object-store bucket. |
Filesystem / Snapshots
Item | Description |
WEKAPP-88400 | Snapshot rename operation is not idempotent Rename a snapshot with existing name is not possible. |
WEKAPP-228616 | Uploading a snapshot to a remote object-store during an object-store detach operation Uploading a snapshot to a remote object-store from a downloaded filesystem while it is undergoing an object-store detach operation can cause cluster unavailability. |
WEKAPP-241668 | Snapshot creation timestamp A snapshot creation timestamp could change to another snapshot's timestamp after a specific sequence of snapshot actions. This issue doesn't affect the content of the snapshots, only the reported timestamp in the snapshot list. |
Filesystem / Squelch
Item | Description |
WEKAPP-213016 | Rare race condition for metadata update during upgrade may lead to data unavailability HCSF has special internal blocks in which it stores a lot of small differences regarding various metadata blocks, in order to reduce both metadata updates and write amplification. These diff-blocks are associated with the addresses of the metadata block database we call a "registry". Portions of the registry, while rare, may split to handle growth of the number of metadata blocks. HCSF version v3.10 introduced new features that required to upgrade these diff blocks. Due to the introduction of this new feature a race condition may occur during the upgrade process of these diff-blocks and during registry splitting due to growth. When this occurs, it can cause diff-blocks to be associated with the incorrect addresses (addresses they had prior to the split). When the diff-blocks are later read (for example, at start-io that happened after the rollback), their addresses undergo a validation process. Due to the race condition, the validation process fails preventing access to the data portion of the blocks. Impact: Data Unavailability |
WEKAPP-227934 | Unavailability after a multi-hop upgrade The system might become unavailable after the upgrade, if there were no modifying IO operations (for example, writes, file creations) on it since the last upgrade. Configuration: Multi-hop upgrade |
Interfaces / Driver
Item | Description |
WEKAPP-227768 | Unable to restart a front-end process If a front-end process is terminated while performing a GPUDirect GDS operation, it cannot be restarted. |
WEKAPP-230629 | Client gets hung on remount When incorrect options are provided to the mount -o remount command during the remount of a filesystem, the system client might get a kernel panic instead of returning a graceful error. |
WEKAPP-232988 | Calling stat immediately after close may report wrong size After a file is written and closed, there is a short window that calling stat() on the file may return the wrong size. |
WEKAPP-234255 | SELinux support with kernels later than 4.5 On kernels later than v4.5, XATTR-NOT-SUPPORTED is wrongly reported as unsupported. Note, to set SELinux, you must mount with -o acl option. |
WEKAPP-234257 | Negative directory entry cache is not invalidated on hardlinks When using a non-zero dentry_max_age_negative mount option, directory entries that become positive due to an hardlink operation do not invalidate negative entries. |
WEKAPP-238861 | Resolved issue with non-4k aligned AIO. Non-4k aligned AIO cause a weka client crash and host reboot. |
WEKAPP-239429 | Resolved issue when using special files on client. The system client generates a Kernel panic under some conditions when using special files (named pipes, block devices) on the system mount point. |
Interfaces / NFS / Ganesha
Item | Description |
WEKAPP-209853 | Unable to mount NFS exports When an NFS export has been created to a non-existing path, once that path is added, users might fail to mount it. The export should be added to an existing path within the filesystem. |
WEKAPP-210775 | NFS metadata performance improvement Significantly improved performance of NFS metadata operations when iterating large directory trees |
WEKAPP-217486 | NFSv3 statistics NFSv3 statistics are not available for NFS-W. |
WEKAPP-218973 | sssd integration with NFS manage-gids option NFS export defined with manage-gids does not retrieve the user's groups information from the configured service in sssd. |
WEKAPP-245771 | Added NFSv3 statistics for NFS-W Added NFSv3 statistics for NFS-W to the GUI and CLI (weka stats commands). |
Interfaces / SMB
Item | Description |
WEKAPP-218971 | Excessive logging to Syslog The SMB and NFS services print excessive logs to Syslog. |
Platform / Network
Item | Description |
WEKAPP-183914 | Potential HA performance impact HA performance might be impacted due to an unbalanced use of both networks |
WEKAPP-206378 | Auto increase MTU Improve MTU increase detection to automatically return to high MTU. |
WEKAPP-217449 | Potential data corruption on RDMA read Under some rare race conditions, when there is a packet loss in the network when working with RDMA, data might get corrupted. Impact: Data corruption upon RDMA read Configuration: RDMA enabled environments |
WEKAPP-220842 | NodeRDMANotActive alert for IO processes that are down NodeRDMANotActive will not be triggered for IO processes that are down and have other indicators. |
WEKAPP-223062 | Clients mount failure The incoming traffic on specified MAC may not route the returned packets correctly. To address this and handle the router HA config properly the configured GW is used instead. |
WEKAPP-225056 | Backend hosts restarts when removing client mounts When the IB network is under congestion and a client is removed there can be a rare race condition that causes some backends to fail with a NetworkPortDead event. Configuration: Infiniband |
WEKAPP-231080 | Unable to remove mixed Infiband and Ethernet networking configuration Unable to remove a mixed Infiband and Ethernet networking on the server. |
WEKAPP-238375 | New QoS statistics Added new statistics to allow a better understanding if a delay is a system side or due to QoS client configuration. WRITE_QOS_DELAY, READ_QOS_DELAY to show the QoS delay and WRITE_LATENCY_NO_QOS, READ_LATEMCY_NO_QOS to show delays without QoS. |
WEKAPP-238399 | Performance hit due to many client restarts Client log files filling the root directory create multiple client process restarts that lead to a performance hit on the cluster (due to too many configuration changes sent by the servers). |
WEKAPP-255458 | Memory leak When the transmitting queues become full, the memory leaks under load. |
WEKAPP-259350 | A process might restart if the network is congested When the network is congested, a process might restart due to a lack of memory buffers (mbufs). |
Platform / SSD
Item | Description |
WEKAPP-257326 | Low efficiency reading A client may reach a state that causes some reads to be performed inefficiently. |
UX / Backbone UI
Item | Description |
WEKAPP-216923 | Authentication preservation during editing of an object-store bucket via the GUI When clicking on the object-store bucket for editing, the authentication method changes to none, instead of preserving the original authentication method. |
WEKAPP-225622 | GUI installation stripe size The stripe size via GUI installation isn't updated to the new limits. For example, on an 8-host cluster, you cannot create a 6+2 (only 4+2) striping via the GUI. The CLI works as expected (for example, allows 6+2 striping with 8- hosts cluster) |
WEKAPP-229456 | Incorrect GUI validation for filesystem minimum capacity The minimum capacity for creating and updating filesystems via the GUI is 10MB while the actual system enforced minimum is 1GB. |
WEKAPP-244172 | Machines Page Synchronization Error message in the GUI After deactivating and deleting a drive, the user cannot add the drive again through the GUI due to a sync issue. A message "Failed Quering Host's Hardware" appears. |
UX / CLI
Item | Description |
WEKAPP-145253 | Resolved Authentication Error Handling Issue Failure to login (weka user login) due to inability to save the auth_token.json file returns success instead of returning an informative error. |
WEKAPP-221844 | CLI output text for REDISTRIBUTING status Changed the text that appears in weka status when the system is redistributing data when it is fully protected (system status is REDISTRIBUTING). |
WEKAPP-232196 | Badly formatted output from the weka alerts command Some alerts were output with bad formatting making the output unreadable. They are now correctly formatted. |
WEKAPP-245797 | Use of weka fs tier [fetch | release | location] in parallel might fail Running weka fs tier fetch, weka fs tier release, or weka fs tier location with more than 16 parallel invocations migth cause internal restarts of system management processes. This will hang and then fail both the invocations and other CLI commands invoked at the time. |
WEKAPP-246209 | CLI login failure exit code A CLI login failure returns 0 (success exit code) instead of 5 (failure exit code). |
WEKAPP-250411 | weka cluster host -v CLI command exits with an error The weka cluster host -v command exits with an error when the cluster includes clients in the DOWN state. Impact: weka cluster host -v with a DOWN client is broken |
WEKAPP-250489 | The login banner message is not displayed The login-banner show CLI command does not display the login banner message. |
UX / React UI
Item | Description |
WEKAPP-265821 | GUI crashes on the statistics page A missing measurement unit in statistics page charts caused the page to crash |
UX / REST API
Item | Description |
WEKAPP-220438 | Filesystem group REST API Added filesystem groups REST API routes. |
Upgrade
Item | Description |
WEKAPP-209545 | Prepare clients for upgrade fails after timeout If the connection has timed out while preparing the clients for the upgrade, the preparation of the clients will fail on rare occasions as the cluster does not keep those connections alive. |
WEKAPP-220843 | Upgrade failure due to incorrect free space validation The free space validation for the upgrade is not accurate and this will result in the upgrade will start even if one of the upgrading hosts doesn't have enough free space, which can result in an upgrade failure. |
WEKAPP-225672 | Upgrade is blocked for a period of time after a cluster leadership restarts Upon a weka management process restart, it appears as if the system comes up from an upgrade. The upgrade task seems to restart its progress from 0% to 100%. During this time, upgrades are blocked. Impact: Cannot upgrade during this time |
WEKAPP-246233 | Upgrade failure An upgrade might fail if an internal configuration change and the cache becomes full. |
WEKAPP-246825 | Reduced the default timeout in upgrade The default timeout during the upgrade process for triggering a rollback is reduced from 5 to 3 minutes. |
WEKAPP-249669 | Upgrade failure The upgrade rarely fails when the system agent fails to parse state files. The failure initiates a rollback to the source version. |
WEKAPP-254470 | Explicit client upgrade failure Running the weka local upgrade command on a client might fail while there are IOs running on the host during the upgrade. |
Upgrade / NDU
Item | Description |
WEKAPP-269693 | Filesystem mount hangs after upgrade A filesystem mount performed right after a weka local upgrade might hang on the client machine due to the system agent failing to wait for front-end processes to finish starting up. |
Known issues
There are no known issues in HCSF v4.0.x.
Documentation
This release includes the following user documentation.
Title | Number and revision | Issue date | |
Hitachi Content Software for File User Guide | MK-HCSF000-01 | October 2022 | |
Hitachi Content Software for File Command Line Reference | MK-HCSF001-01 | October 2022 |
Copyrights and licenses
© 2022 Hitachi Vantara LLC. All rights reserved.
No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including copying and recording, or stored in a database or retrieval system for commercial purposes without the express written permission of Hitachi, Ltd., or Hitachi Vantara LLC (collectively "Hitachi"). Licensee may make copies of the Materials provided that any such copy is: (i) created as an essential step in utilization of the Software as licensed and is used in no other manner; or (ii) used for archival purposes. Licensee may not make any other copies of the Materials. "Materials" mean text, data, photographs, graphics, audio, video and documents.
Hitachi reserves the right to make changes to this Material at any time without notice and assumes no responsibility for its use. The Materials contain the most current information available at the time of publication.
Some of the features described in the Materials might not be currently available. Refer to the most recent product announcement for information about feature and product availability, or contact Hitachi Vantara LLC at https://support.hitachivantara.com/e...ontact-us.html.
Notice: Hitachi products and services can be ordered only under the terms and conditions of the applicable Hitachi agreements. The use of Hitachi products is governed by the terms of your agreements with Hitachi Vantara LLC.
By using this software, you agree that you are responsible for:
1) Acquiring the relevant consents as may be required under local privacy laws or otherwise from authorized employees and other individuals; and
2) Verifying that your data continues to be held, retrieved, deleted, or otherwise processed in accordance with relevant laws.
Notice on Export Controls. The technical data and technology inherent in this Document may be subject to U.S. export control laws, including the U.S. Export Administration Act and its associated regulations, and may be subject to export or import regulations in other countries. Reader agrees to comply strictly with all such regulations and acknowledges that Reader has the responsibility to obtain licenses to export, re-export, or import the Document and any Compliant Products.
Hitachi and Lumada are trademarks or registered trademarks of Hitachi, Ltd., in the United States and other countries.
AIX, AS/400e, DB2, Domino, DS6000, DS8000, Enterprise Storage Server, eServer, FICON, FlashCopy, GDPS, HyperSwap, IBM, Lotus, MVS, OS/390, PowerHA, PowerPC, RS/6000, S/390, System z9, System z10, Tivoli, z/OS, z9, z10, z13, z14, z/VM, and z/VSE are registered trademarks or trademarks of International Business Machines Corporation.
Active Directory, ActiveX, Bing, Excel, Hyper-V, Internet Explorer, the Internet Explorer logo, Microsoft, Microsoft Edge, the Microsoft corporate logo, the Microsoft Edge logo, MS-DOS, Outlook, PowerPoint, SharePoint, Silverlight, SmartScreen, SQL Server, Visual Basic, Visual C++, Visual Studio, Windows, the Windows logo, Windows Azure, Windows PowerShell, Windows Server, the Windows start button, and Windows Vista are registered trademarks or trademarks of Microsoft Corporation. Microsoft product screen shots are reprinted with permission from Microsoft Corporation.
All other trademarks, service marks, and company names in this document or website are properties of their respective owners.
Copyright and license information for third-party and open source software used in Hitachi Vantara products can be found in the product documentation, at https://www.hitachivantara.com/en-us...any/legal.html or https://knowledge.hitachivantara.com...ource_Software.