P5 Release Notes
New Features in Archiware P5 Version 7.1 (October 4, 2022)
- P5 Data Mover: an extension of the P5 Archive module, which offers policy-based moving and copying of archived data between tape, cloud and disk.
- Multi-factor authentication: provides an added layer of security.
- MHL checksum manifest of LTFS archived data: ensures that the source data has been saved and recovered without any alterations.
- UNC path support: enables easy access to network source and target storage.
- Improved database design: The data model for the archive and backup indexes has been revised and simplified by removing tables and indexes that were necessary in the past for compatibility with older versions. This reduction in complexity provides greater security and scalability.
- Browser-GUI improvements for large/complex installations: Thanks to improved database design, new query functions are possible that are particularly effective for very large P5 installations containing many objects. For example, with installations containing several thousand LTO volumes, admin tasks in the browser are even quicker and more efficient than was previously possible.
- Metadata and proxy/preview import from third-party MAM/DAM/PAM into archive index for existing archive files
Previous Release Notes
Version 7.1.7 released. Changes since last release:
- Fixed upload to cloud buckets with default object lock set.
- Workarround handling of "ALL" placeholder in backup task.
- Fixed calculation of up/dowload streams for objectstore access.
- Fixed container volume used size after migration. (Bug#1816)
- Fixed starting Sync jobs from the CLI. (Bug#1849)
- Improved restore collection speed for very large indexes. (Bug#1855)
- Added utility to fix missing volume clone index records. (Bug#1857)
- Improved handling for container staging area overfill.
Version 7.1.6 released. Changes since last release:
- Fixed linking cycle paths when longer than 256 chars on Windows. (Bug#1094, Bug#1464)
- Fixed Sync on Windows for UNC paths. (Bug#1699)
- Skip disabled volumes from all migration operations. (Bug#1743)
- Fixed migration of imported LTFS volumes - missing date written. (Bug#1737)
- Fixed verify for volume-spanned files on LTFS volumes. (Bug#1759)
- Fixed restoring migrated files. (Bug#1752)
- Fixed REST PoolCreate call lacking CONTAINER type. (Bug#1748)
- Fixed backup index save for indexes > 4GB on Windows. (Bug#1741)
- Fixed partial restores. (Bug#1676)
- Avoid importing wrong time stamps from foreign LTFS indexes (Bug#1782)
- Improved handling of disabled volumes during restore/migration.
- Fixed calculation of blocked volumes for progressive backups.
- CLI volume-based restore uses original paths. (FR#1776)
- Fixed "invalid argument" problems that appeared for some SAN/NAS mounted file systems. (Bug#1756)
- Added support for SNI for S3/compatible cloud providers.
Version 7.1.5 released. Changes since last release:
- Fixed further problem with import of Xendata LTFS with with large index generation numbers. (Bug#1664)
- Fixed problem with reading streams on Windows. (Bug#1669)
- Fixed usage of interpunction characters in path passed as alternate archive index root. (Bug#1665)
- Fixed size of imported LTFS volumes. (Bug#1697)
- Fixed random "database is locked" errors during lookup table creation in preparation for backup. (Bug#1653)
- Fixed occasional incomplete restores. (Bug#1676)
- Allow parallel migration for P5-volumes. (Bug#1689)
- Fixed container management for extremely short retention times and occasional backup jobs. (Bug#1681)
- Invalidate volume-clone records during move-migration.
- Fixed "file truncated" problems on Windows. (Bug#1710)
- Fixed volume imports to avoid duplicates. (Bug#1713)
- Fixed index-recover from VTL's and regular P5 tape volumes.
- Fixed save to LTFS on Windows of files with offending and special characters in file names. (Bug#1731)
- Fixed access right problems for files on Windows. (Bug#1684)
Version 7.1.4 released. Changes since last release:
- Fixed file deletion after archive. (Bug#1639)
- Fixed inode optimization for backup.
- Fixed recover volume from clone. (Bug#1633)
- Improved container management w.r.t. backup of moved files.
- Fixed restore for files missing index path.
- Avoid descending into __CYCLES when linking new sync cycle.
- Fixed migration report about required volumes w.r.t. size/count of files to be migrated. (Bug#1638)
- Fixed restore from an UNC path that led to infinite loop. (Bug#1628)
- Added [cli::index::connectdirs] utility.
- Added [cli::index::prunedirs] utility.
- Fixed Windows preview generation for UNC paths. (Bug#1654)
- Fixed Sync with Verify on Windows. (Bug#1367)
- Fixed block transcoding/crash for Synthetic Backup or DataMover.
- Fixed saving sparse files on Windows. (Bug#1657)
- Worked around import of improperly formatted LTFS tapes originating from Xendata. (Bug#1664)
- Fixed dangling CLI sessions that prevented restarts. (Bug#1648)
Version 7.1.3 released. Changes since last release:
- Added some CLI index/volume management calls.
- Fixed rare cases where increment would save too much. (Bug#1620)
- Improved cleanup of expired containers.
- Improved handling of old file versions in backup.
- Fixed occasional file-size calculation problems on Windows. (Bug#1506)
- Fixed simultaneous restore from container and other media.
- Backup index purge now always purges elements and addresses.
- Added MiB/GiB/TiB and friends.
- Fixed broken B2Go server connections. (Bug#1630)
- Fixed import volume import that occasionally produced duplicate file and folder entries in the index. (Bug#1623 Bug#1631 Bug#1632)
- Fixed disaster recovery from container/cloud backup. (Bug#1626)
- Fixed File framing error and read over upper boundary during restore on Windows. (Bug#1627)
Version 7.1.2 released. Changes since last release:
- Fixed backup of root (/) filesystem.
- Fixed backup/archive to containers from remote clients.
- Better handle session expiry for long running CLI commands.
- Corrected handling of snapshot time during restore (Bug#1617)
- Fixed restore of files saved on tape media to remote clients having interleaved blocks (i.e. saved using parallel streams).
- Adjusted timeout for saving of backup indexes to object stores.
- Fixed restore of spanned LTFS files.
Version 7.1.1 released. Changes since last release:
- Fixed LTFS archive of filenames beginning with a tilde. (Bug#1588)
- Added forceExpiryTime to force progressive strategy for VTL volumes that are mirrored on the cloud.
- Added forceFilewalkerTwoPass to force walking the filesystem twice, first time to collect all new/changed files, second time to collect files that need be unconditionally saved.
- Fixed config DB lookup for description key. (Bug#1590)
- Fixed (rare) case when initial backup may save too much when applying some seldom used hard-wired config options.
- Fixed restore of archived items stored directly on the cloud (not within our container or VTL). (Bug#1592)
- Fixed LTFS error EDABGEN on Windows. (Bug#1599)
- Fixed LTFS writing/flushing index at EOM. (Bug#1585)
- Fixed restore of excess files to remote clients. (Bug#1601,1600)
- Fixed (rare) index writer thread errors. (Bug#1606,1598)
- Fixed rest errors. (Bug#1594,1595,1596)
- Fixed compatibility with pre-7.1.1 clients. (Bug#1604)
Version 7.1 released.
For new features, please refer to the 7.1 Features List.
Version 7.0.7 released. Changes since last release:
- Propagates DYLD_LIBRARY_PATH env setting to LaunchDaemon setup on macOS.
- Fixed handling for S3 Intelligent Tiering.
- Removed media volume block sizes > 512K.
- Fixed verify of zero-length files stored on LTFS.
- Fixed auto-restart during delegated syncs.
- Added (unsupported) option for handlig data-forks only.
- Stringent check for backup plan ID during container backup.
- Cleanup expired containers with all refd. files deleted.
- Fixed broken synthetic save.
- Selection for restore avoids creating large temp tables.
Version 7.0.6 released. Changes since last release:
- Included binaries for macOS M1 port.
- Allow cancel of stalled cloud uploads after 1hr timeout.
- Fixed artificial tape EOM.
- Fixed rare crash while using VTL volumes.
- Added pro-active save of data located in sparsely filled containers (with less than 20% of active files).
- Improved cancellation of jobs on VTL volumes.
- Fixed potential restore loop for client restores.
- Removed 60-chars limitation on mail server passwords.
- Fixed removal of stale addresses for deleted containers.
- Fixed login credentials encryption for Google ObjectStore.
- Added LZ4 compression for VTL volumes.
- Removed content-based verification for archive jobs.
- Fixed restore from LTFS volumes to remote clients.
- Fixed loading Helios Desktop Stub for Base V6.0.0
Version 7.0.5 released. Changes since last release:
- Fixed LTFS verify (Bug#1439).
- Avoid re-partitioning of LTFS tapes if existing partitions are large enough (fix for support of upcoming LTO9 tapes).
- Added multi-stream download from Cloud object stores.
- Fixed filter usage for progressive strategies (Bug#1446).
- Fixed LTFS handling of splitted files (Bug#1454).
- Extended index backups to maintain file history for at least 6 months. Not exposed over GUI yet.
- Fixed container database handling for very long backup jobs (adds more frequent check-points).
- Fixed automount of RDX drives (Bug#1460).
- Fixed possible hang in periodic LTFS index flush (Bug#1456).
- Fixed import of LTFS tapes containing file-names that start with tilde-character (Bug#1459).
- Fixed endless loop with Glacier uploads.
- Added CLI/REST commands for changing client access credentials.
- Added CLI/REST commands to fully manage backups create/edit/schedule/delete plans, tasks, schedules.
- Added CLI/REST commands to create/edit/delete filters.
- Fixed volume migration for Disk -> Tape (Bug#1462).
- Disabled content-verify for archive jobs altogether due to the incompatibility with file/ctime handling on Windows.
- Added check for space left on tape to overcome the possibly skipped early warning EOM mark that has lately been observed on some instances of the IBM LTO firmwares. (Bug #1458).
- Fixed Sync with Cycles on Windows.
Version 7.0.4 released. Changes since last release:
- Critical bug: Volume(s) recycling causes excessive index cleanup. This bug has been fixed.
- Fixed calling Windows exec proxy and preview generation via the proxy.
- Added verify option to cloud drive that performs automatic verify-after-write.
- Increased restore robustness for restores from damaged container files.
- Fixed dry-run start for backup plans with more then one registered task.
Version 7.03 released. Changes since last release:
- Enforce time-skew between archive/backup jobs even when jobs are distributed over clients.
- Fix container bigfile handling when a backup of a bigfile fails prematurely w/o writing a single block of data.
- Fixed FreeBSD xattr treatment during file reading that may lead to crash under circumstances.
- Fixed progressive backup of remote clients by preventing auto-expiry of remote file-walker.
- Fixed deadlock while spanning LTFS tapes.
- Fixed server uptime timer for Linux/Solaris/FreeBSD.
- Fixed restart-times parser to allow octal-like numbers.
- Fixed preview generator for ffmpeg to allow absolute executable paths.
- Fixed GUI-based P5 update for Linux/FreeBSD/Solaris.
- Improved CLI to allow "now" as submit option.
- Removed LTFS verbosity in log.
- Fixed nsdchat for Windows
- Better control of server restarts for client operations.
Version 7.0.2 released. Changes since last release:
- Avoid wholesale owner-change on large directories during startup.
- Fixed device-manager related problems on Windows.
- Improved update from older P5 releases.
- Fixed GUI-based P5 update for Mac.
Version 7.0.1 released. Changes since last release:
- Fixed media blocksize seed for remote clients.
- Fixed LTFS attributes for repeatedly archived files.
Version 7.0 released.
Version 6.1.2 released. Changes since last release:
- Fixed backward-compatibility with some user-defined callbacks.
- Added Dry-Run capability for backup runs. For the moment as a manual option.
- Recompiled speed-sensitive code for internal MD5 generation will speed-up archive operations on Windows. (Bug#1330)
- Cloud volume sync is now run also for jobs that have been time-limited.
- Fixed regenerate from clone when tapes were written or labeled with different P5 versions. (Bug#475, #1267)
- Enabled moving and linking the P5 config directory into a different location or to a separate file system. (Bug#1239)
- Fixed allowing to delete an orphaned drive when its library got deleted. (Bug#1268)
- Fixed broken ZIB compressor or Windows. (Bug#1274)
- Fixed potential danger of overwriting tapes in a single drive when the job is waiting for a new empty tape and is then interrupted. (Bug#1284)
- Resolved occasional protocol errors occurring when transferring data to Backblaze. (Bug#1298)
- Fixed REST interface restore selection problem. (Bug#1301)
- Fixed problems with Import of backup volumes. (Bug#1316)
- Fixed diverse problems with file names and the LTFS index. (Bug#1320)
- Fixed problem at startup while calculating volume predecessors. (Bug#1338)
- Improved detection of ffmpeg on Windows and corrected passed command line options. (Bug#1350)
Version 6.1 released.
For new features, please refer to the 6.1 Feature List.
Version 6.0.2 released. Changes since last release:
- Fixed update of user/group records during V6 transition (Bug#1220).
- Fixed autostart on system boot for CentOS7 (and later) (Bug#1223).
- Fixed error/warning report for B2Go jobs running against missing or empty top-level directories (Bug#1235).
- Fixed B2Go configuration over CLI to gracefully handle multiple server records (Bug#1246).
- Deleting the client also deletes its filesystem DB (Bug#1247).
- Fixed backup task config hint about ignoring filesystem boundary during file/index scan (Bug#1258).
- Added CLI [User password] command to manipulate GUI login passwords.
- Added CLI [RestoreSelection addfromvolume] to be able to restore all files located on the given volume.
- Improved CLI [resolvedups] index maintenance command to handle duplicates in the paths table.
- Added ns/servers/lexxsrv/module/Sync/cycleCreateIgnoreErrors config option to ignore file-related errors while hard-linking
- Fixed saving file systems mounted at mountpoints on Windows. (Bug#1234, Bug#1253)
- Fixed job scheduling issues / jobs hanging after start in Windows. (Bug#1238)
- Fixed P5 process terminating without any apparent cause on Windows. (Bug#1221)
Version 6.0.1 released. Changes since last release:
- Notarized for macOS Catalina.
- Added missing application page for bulk volume delete (Bug#1176).
- Fixed typo in start-server script.
- Fixed (re-)creation of ACLs on restore for Mac builds (Bug#1181).
- Fixed setting group path references for many clients (Bug#1187).
- Fixed installing regular license for Desktop edition (Bug#1185).
- Fixed user check in verify/restore to remote clients (Bug#1184).
- Fixed handling of some illegal chars in REST interface (Bug#1177).
- Fixed turning-off data compression (Bug#1175).
- Fixed unzip of restore selection during B2Go restore (Bug#1174).
- Fixed spelling in B2Go log (Bug##1173).
- Fixed handling of restore over restore selection when multiple snapshot have been selected (Bug#1172).
- Fixed handling configuration DB on Windows (Bug#1162).
- Object archives now support only non-empty regular files (Bug#1160).
- Fixed auto-schedule of B2Go workstations (Bug#1158).
- Added hint about HTTP protoocol port in client config (Bug#1149).
- Fixed upload to Google cloud storage (Bug#1145).
- Fixed restore selection that restore same file(s) twice when located on two/more volumes (Bug#1140).
- Fixed auto-deletion of old LTFS-pools (Bug#1134).
- Fixed restore from Sync cycles on BTRFS (Bug#1049).
- Fixed support for passwords with embedded colon (Bug#932).
Version 6.0 released.
For new features, please refer to the 6.0 Features List.
Version 5.6.6 released. Changes since last release:
- Fixed archive job inventory for local/remote clients. (Bug #1100)
- Fixed protection issues during archive verify. (Bug#778)
- Fixed rare crashes on Windows possibly involving jukebox heartbeat monitor. (Bug #1126)
- Fixed Windows restore for existing directories. (Bug#1124)
- Fixed Windows crash during Backup on ReFS. (Bug#1099)
- Fixed auto-recognition of Qnap devices. (Bug#1102)
- Fixed general slowdown of filesystem operations on Windows.
- Storage devices now re-check block position after having written 4096 blocks in a row.
Version 5.6.5 released. Changes since last release:
- Fixed rename/delete of user-forks for Helios.
- Fixed manual archive index purge (Bug#1021).
- Allow operator users to stop/start/cancel jobs(Bug#1037).
- Disable legacy options for archive preview generators based on Apple QuickTime (Bug#1038).
- Fixed archive index backups on filesystems that do not support xattr (Bug#1042).
- Fixed object store uploads that crashed while reporting invalid object ID (Bug#1046).
- Fixed delegated client2local sync (Bug#1064).
- Fixed email attachment for non-localhost clients (Bug#1066, Bug#1067).
- Fixed "permissions denied" sync case on Windows (Bug#1076).
Version 5.6.4 released. Changes since last release:
- Fixed handling of symbolic links for files on Windows when synced from a different OS. (Bug#997)
- Fixed "not null" constraints error when filling the filesystem table during the backup. (Bug#994)
- Fixed storage of Mac decomposed-unicode files on foreign (non Mac) filesystems. (Bug#988).
- Fixed CLI "ArchiveSelection destroy" behaviour to return correct values. (Bug#984)
- Fixed field-name in B2 cloud template. (Bug#982)
- Fixed rare crash issues during cloud upload. (Bug#978)
- Fixed error during folder-creation out of the file-picker on Windows. (Bug#977)
- Fixed TCC issues for Mac Mojave OS. (Bug#972).
- Changed/fixed browser link for Qnap NAS
- Changed naviserver port from 8000 to 8001 on Qnap NAS
- Work-around problems of chunked cloud volumes where positively acknowledged file/chunk upload does not result in the chunk being retrievable from the service. For additional security we now double-check the operation by initiating a restore attempt for the same chunk.
Version 5.6.3 released. Changes since last release:
- Fixed mail re-send logic to regenerate temporary files after each try. Also fixed awmail to correctly generate sub-second unique message IDs (Bug#945).
- Fixed recognition of filesystem volumes located deeper in the directory structure (Bug#954).
- Added new CLI command [RestoreSelection size] which returns total size in bytes of all files in the selection object (FR#950).
- Declared CLI [ArchiveSelection size] deprecated, use [ArchiveSelection entries] instead.
- Improved CLI selection handling to allow multiple parallel selections open at any time.
- Fixed typo in this file for 5.6.2 release that references ns_section commands (Bug#946).
- Removed drive-type check for LTFS (Bug#953).
- Fixed support for 2+ simultaneously mounted LTFS volumes (Bug#958).
- Added config database optimisations to reduce concurrency and number of updates for certain job types (Bug#959).
- Improved error reporting for client2client sync.
- Fixed backing up archive index on Windows (Bug#951).
- Fixed tape drive cleaning that was broken in 5.6.2. (Bug#914)
- Fixed remote update for FreeBSD clients (Bug#957).
- Fixed CLI creation of indexes (Bug#965).
- Fixed timeouts in Client2Client syncs when large source datasets were involved (Bug#970).
Version 5.6.2 released. Changes since last release:
- Re-introduced FreeBSD/amd64 port for FreeBSD V11.
- Deprecated: SCSI device support.
- Re-introduced Solaris/Intel port for Solaris 10+.
- Deprecated: SCSI device support.
- FR#930: Adding metadata in P5 archive.
- Added more clever logic: per-job metadata can be overriden by non-empty per-file metadata. One can mix both but one cannot combine/merge values.
- FR#929: Backup of nested filesystems.
- In order to support backup of large/deep hierarchicalfilesystems (like for example ZFS) one can now enablecrossing filesystem boundary while walking the filesystem tree and looking for changed files. To do that, one needs to edit config/lexxsrv.<PORT> file (where <PORT> is the value assigned during the installation; default = 8000), and put those two lines somewhere at the end of the file:
- ns_section ns/server/$server/module/Backup/CLIENT
- ns_param crossFilesystemBoundary true
- where CLIENT is the name of the P5 client.
- Default value for crossFilesystemBoundary is "false", which means that file walk will not cross filesystem boundary.
- After changing the file, P5 needs to be restarted for changes to become active.
- FR#895: volumes with backing cloud-stores can now be manually synced-up with the cloud storage, regardles of their internal state. This aids in cleaning-up the volume cache for volume data stored in cloud only.
- FR#875: sending e-mail is now more persistent in case of mail-server connection problems.
- FR#874: added CLI command "Volume jobs" that lists job names of all the jobs that were writing to the volume.
- FR#861: added tunable size for disk IO operations.
- To change disk IO size, edit the config/lexxsrv.<PORT> file (where <PORT> is the value assigned during the installation; default = 8000) and put those two lines somewhere at the end of the file:
- ns_section ns/server/$server/module/jobex/CLIENT
- ns_param bioSize SIZE
- where CLIENT is the name of the P5 client and SIZE is a value in bytes, aligned to 512.
- After changing the file, P5 needs to be restarted for changes to become active.
- Added multithreading support for Sync cycle creation on filesystems w/o snapshot option.
- To enable MT, one needs to edit config/lexxsrv.<PORT> file (where <PORT> is the value assigned during the installation; default = 8000), and put those two lines somewhere at the end of the file:
- ns_section ns/server/$server/module/Sync
- ns_param cycleThreadCount 1
- Default value for cycleThreadCount is 1 (one). Increasing number of threads for cycle generation might speed up this (lenghtly) operation if target filesystem is located on SSD device or on a very fast RAID system.
- After changing the file, P5 needs to be restarted for changes to become active.
- Fixed B2Go connection losses after 3 minutes waiting for the next changed file (Bug#928).
- Fixed entering old-style licenses (Bug#925).
- Fixed timeouts in Sync during long snapshot creation (Bug#923).
- Fixed mail delivery when two or more e-mails were sent within one second interval (Bug#921).
- Fixed cleanup of backup indexes containing large number of recycled volumes (Bug#918).
- Fixed handling of Mac resource forks for HFS+ -> HFS+ copies (Bug#915).
- Added "ignore cleaning flag" for special applications.
- To ignore the cleaning flag for selected drives, edit the config/lexxsrv.<PORT> file (where <PORT> is the value assigned during the installation; default = 8000) and put those two lines somewhere at the end of the file:
- ns_section ns/server/$server/module/medmgr
- ns_param dontClean "DEV1 DEV2 ... DEVn"
- where DEV1, DEV2, etc. are names of the tape devices as shown in the P5 GUI.
- After changing the file, P5 needs to be restarted for changes to become active.
- Fixed write-speed issues for large-file archives (Bug#913).
- Fixed auto-recognition of EtherShare volumes (Bug#907).
- Fixed crashes on archives to cloud when jobs terminate unexpectedly or get cancelled manually (Bug#904).
- Fixed crashes on Mac (and possibly other platforms) when rebuilding indexes after the backup.
- Reduce number of warnings in GUI about Sync plans running in transfer mode (Bug#900).
- Fixed restore entry for Sync plans when there is only one Sync license configured (Bug#989).
- Fixed mode of resource sub-directories created during the restore within EtherShare volumes to 0777 (Bug#892).
- Fixed spurious virtual tape driver errors claiming the volume being written has no label attached (Bug#891).
- Fixed loading the LTFS driver for IBM LTO drives (Bug#890).
- Fixed Sync file-event handling when source of events is not the localhost computer (Bug#888).
- Fixed "Back Arrow" side-effects in GUI after log-in (Bug#885).
- Added missing Client2Client manual Sync option (Bug#884).
- Removed Cancel button from log-out pane (Bug#882).
- Fixed Backup job lock-out after Volume recycle (Bug#877).
- Added Application-Key support for BackBlaze cloud (Bug#858).
- Fixed resource fork handling during restore to non-HFS filesystems resulting in IO-error messages (Bug#901).
- Fixed generic install script on Unix to ignore Synology install options (Bug#842).
- Removed scanning files before archive and backup. This feature was simply consuming resources and did not provide any really useful information.
- Fixed label of cloud volumes in case of upload errors (Bug#820).
- Improved bandwidth throttle code to handle 10Gbps ethernet links (Bug#143).
- Removed extended log from job summary
Version 5.6.1 released. Changes since last release:
- Fixed issues when logging-in as root into the GUI.
- Fixed special characters handling in Sync from Helios (possibly also Xinet) fileserver volumes.
- Introduced new network throttle implementation.
- Fixed multi-part upload (1GB+ files) for AWS cloud.
- Fixed auto-start on boot for Linux Ubuntu 18+ version.
- Fixed Sync job handling that caused jobs to stick in the job queue until inspected by the GUI.
Version 5.6 released.
For new features, please refer to the 5.6 Features List.
Version 5.5.6 released. Changes since last release:
- Fixed device handling for incremental backups that caused unnecessary full backups sometimes.
- Added log about used archive index offset to aid support.
- Fixed snapshot purge for backup that caused far too many snapshots to be held online.
- Fixed treatment of user preference records with embedded blanks.
- Fixed edge case where we could delete the volume chunk from the cloud that is not cloned locally.
Version 5.5.5 released. Changes since last release:
- Fixed handling for BackBlaze file deletions.
- Fixed core when backing up B2Go repository.
- Fixed import of volumes.
- Fixed preview generation that got broken with the recent fixes in 5.5.4 w.r.t file uploads
- Fixed reading resource fork on Windows after being broken by Windows security update.
Version 5.5.4 released. Changes since last release:
- Improved file upload/download to/from clients. Uses timed non-blocking transfer with regular http post/get sequence. Previous behaviour resulted in partial transfers on heavily loaded MT-machines. Transfer is still limited to 2GB.
- Suppressed creation of void database files outside the config directory (Bug #778, #775).
- Added object store volume cleanup job at the end of executing a backup/archive job.
- Fixed handling of bonded file-systems for backup.
- Fixed race condition for CLI jobs that reported completed state before actually started by the scheduler thread (Bug #776).
- Added a checksum to P5 exe on Windows to accomodate the Tandberg RDX Access Control client and the ransom blocker.
- Fixed file-events handling on Mac to cope with last-reported-event of zero which may result in irregular overflow errors.
- Changed ID's of the objects stored on object stores to look like paths so implementations may optimize.
Version 5.5.3 released. Changes since last release:
- Fixed QNAP creation of folders from the GUI file selector.
- Fixed case when two or more archive jobs work on the partially overlapping directory structures that have never been archived before (Bug #684/#753).
- Fixed ctime disable setup in config file (Bug #731).
- Fixed Backup2Go restore on Windows clients (Bug #741/#742).
- Fixed media changer internal error when attempting to mount subsequent volume(s) (Bug #751).
- Fixed volume regeneration in disk media changer (Bug #756).
- Fixed CLI to properly check username for socket conns (Bug #757).
- Fixed CLI inventory commands that were not reporting entry handles at all after update to V5.5 (Bug #759/#764).
- Fixed file-events sync that was causing too much directories to be reported as changed.
- Fixed cases where we hit locked database on various index ops.
- Added basic support for new Apple APFS filesystem (FR #744).
Version 5.5.2 released. Changes since last release:
- Added LTO-8 to the list of supported tape drives.
- Added separate config file options for testing inode change time for backup and archive jobs and combined.
- Fixed client2client sync that occasionally left the synced files on the server machine.
- Fixed stop signal for archive/backup jobs to correctly propagate to object store up/downloaders.
- Fixed B2Go encryption backup for (at least) Mac clients.
- Fixed object store up/downloaders on Synology devices that were occasionally stalled on an unknown file descriptor.
- Fixed access to Manual Archiving for non-privileged users who are granted archiving privilege.
Version 5.5.1 released. Changes since last release:
- Fixed treatment of user-defined meta keys for archive indexes which contain interpunction characters (Bug#701).
- Fixed volume-chunk deletion during volume re-label (Bug#702).
- Added option to configure time checking separately for archive and backup incremental operations (FR#706).
- Fixed bad connection test for Generic S3 cloud services. Also, reduced timeout on connection problems from five to one minute, as not to block GUI for too long (Bug#708).
- Fixed some minor compatibility issues in generic S3 object store connector.
- Improved remote agent login handling for delegated syncs.
- Fixed occasions where regular blockpool teardown might be interpreted as error, effectively making valid jobs to fail.
- Fixed occasional index corruption happening after some index maintenance tasks.
- Added cloud wizard in the GUI.
- Fixed occasional communication issues in Sync jobs when transferring very small checksum difference file(s).
- Honour job stop signals within volume chunk uploads.
- Fixed B2Go encryption handling for backup.
- Fixed job monitor issues where a cascade of running jobs may miss the top-level job slot for progress/cancel.
Version 5.5 released.
For new features, please refer to the 5.5 Features List.
Version 5.4.4 released. Changes since last release:
- Updated CLI with [Workstation name] command, that when executed on the workstation, will return its name/ID assigned on the server (FR#34).
- Updated CLI [Volume location] to show info about the slot if the volume is found to be in a media changer (FR#570).
- Fixed wire-protocol error when restoring multi-volume job to remote clients (Bug#645.
- Fixed volume usetime and usecount during tape recycle (Bug#641).
- Fixed job result after restore job ran in error (Bug#632).
- Fixed B2Go file deletion of files containing special escaped characters, that left the file hanging in repository (Bug#628).
- Suppresses CLI Archive jobs with empty archive selection. The submit command will simply return error (Bug#626).
- Fixed README1st.txt in Windows distribution (Bug#625).
- Fixed incremental backup that turned to full after device ID changes (Bug#621).
- Set configuration database file to be read by root only (Bug#614).
- Fixed automatic B2Go workstation upgrade (Bug#613).
- Fixed cases where verify after archive required additional media volumes that were not used by the job (Bug#610).
- Fixed Sync job failure when syncing to Helios Volumes with cycles set (Bug#600).
- Fixed nsdchat to recognize URL-encoding in awsock/awfile targets, as to be able to pass "@" (and perhaps other special characters) in user/password fields (Bug#599).
- Fixed incremental backup for re-mounted Volumes (Bug#589).
- Changed shell-scripts to include "P5" product name (Bug#563).
- Fixed processing of "restore-to" parameter in RestoreSelection addentry CLI command (Bug#492).
- CLI commands for BackupPlan start/stop are now no-ops.
- Fixed BTRFS snapshot creation when target directory is on a subvolume (FR#652).
- Fixed CLI Job status command that may return wrong results due to timing issues (Bug#658).
Version 5.4.3 released. Changes since last release:
- Fixed switching of snapshots w.r.t. GUI browser position.
- Added [ArchivePlan incrlevel] CLI command.
- Added [Pool enable|enabled|disable|disabled] CLI commands.
- During restore, recognize admin users as privileged.
- Added cli::index::fixmedia index utility command.
- Fixed server-restore of B2Go workstation data.
- Improved cancellation of cloud archive jobs.
- Fixed excessive CPU usage for cloud archive jobs.
- Fixed encrypted-data backup/archive from clients.
- Fixed volume migration and synthetic backup.
Version 5.4.2 released. Changes since last release:
- Improved object store up/download handling by reducing log calculation frequency and thread creation/teardown.
- Fixed sole-file CLI restores in the compatibility mode.
- Produce better error log when trying to restore encrypted files to client w/o correct encryption key set.
- Fixed incremental archive for object store archives.
- Filter out .zfs directory from being operated on.
- Skip disabled and/or P4 archive indexes from the object store usage calculation.
- Fixed call to Windows user-defined preview scripts.
- Fixed preview generation for some oddly named files.
- Fixed restore to remote clients when hitting the filemark.
Version 5.4.1 released. Changes since last release:
- Fixed broken Sparc/64bit port.
- Skip auto-update for clients that are servers themselves.
- Fixed indexer error when adding cloned address records.
- Fixed rare media-block formatting error for checksum trailers.
Version 5.4 released.
For new features, please refer to the 5.4 Features List.
Version 5.3.4 released. Changes since last release:
- Improved handling of hierarchically mounted filesystems during restore.
- Improved interaction with FreeBSD CAM system.
- Fixed crashes related to media changer heartbeat thread when running on FreeBSD hosts.
- Fixed erroneously setting old volume to suspect after recycling.
Version 5.3.3 released. Changes since last release:
- Added support for Synology DSM platform
- Workaround for Overland NEO drive serial number reporting while querying the library for element status.
- Corrected time stamp on restored and synced folders on Windows 64-bit.
- CLI now reports all jobs that were either scheduled, started or running as "terminated" if the server was pulled down or crashed during the operation. (Bug #458)
- Fixed index behaviour for directories that lost their address record somehow (perhaps during volume deletions). (Bug #444)
- Fixed server crash during verify runs on archive if the client encryption was turned on. (Bug #447)
- Fixed "invalid-id-to-filename" errors when restoring to clients if the client encryption was turned on. (Bug #457)
- Fixed backup of empty top level directories. (Bug #462)
- Fixed nsdchat utility interactive mode not to block on "exit" command. (Bug #121)
- Fixed index handling when 2 Full backup jobs operate on same directory structure and are executing "near" each other, timewise. (Bug #464).
- Fixed sync in update mode when remote file versions are enabled and source contains hard-linked files. (Bug #421)
- Added cli::index::prunedirs method to clear the index from empty directories not found on any known volummes or ones created as helping stepstones for maintaning tree integrity.
- Fixed delete-stamp of files in the index that have been deleted on the filesystem to match the backup time of the increment run that found them missing. (Bug #474)
- Fixed expression for locating changed files for incr Archives. (Bug #471)
- Improved index re-page routine to backup the live DB file before the attempt to re-page and/or filter the records.
- Added option to set port 587 with default TLS protocol for mail delivery. The port 465 is reserved for the SSL.
- Fixed incremental archive for directories that have been moved or renamed in the index.
Version 5.3.2 released. Changes since last release:
- CLI restores now corectly select the right timed-instance of the file at the given path.
- Fixed situations in Archive index when same paths are being moved and re-archived again.
- Fixed mail delivery to better handle various security-related mailer setups.
- Fixed finder-info restores on Xinet volumes when files are being relocated into completely empty directories.
- Fixed handling of parametrized preview and/or metadata user-defined scripts for archive jobs.
- Fixed handling of clone volumes where we could require both the volume and its clone during the verify passes.
- Skip generating previews where internal preview generator was implicitly disabled and the provided user-defined preview generator decied not to make previews for certain file types.
- Fixed detection of Linux devices for high LUNs and for DDP.
- Fixed search GUI on Windows.
- Fixed recognition of non-Mac streams within Helios volumes.
- Corrected test mail sending when the "New Notification" event was removed from the list of configured events.
- Archive job inventory file is now automatically ZIP'ed if larger then some predefined value. The default size above which the file is zipped is 16MB. This can be controlled by the config option:
- ns_section ns/server/$server/module/job
- ns_param maxInventorySize 16777216
- Removed traces of Growl from the notifier code.
- Fixed rare race-condition in indexer for backup operation related to backups with enabled file-prescan.
Version 5.3.1 released. Changes since last release:
- Fixed linking to correct binaries on new installs.
- Uses /sys instead of /proc when looking up SCSI devices on newer Linux installs.
- Simplified and improved GUI interface for search function in Archive and Backup indexes.
- Fixed: 64-bit Solaris11 SCSI device access.
Version 5.3 released.
For new features, please refer to the 5.3 Features List.
Version 5.2.2 released. Changes since last release:
- Fixed lookup selection when generating entries for root.
- Fixed query expression when selecting/sorting nodes on backup time column.
- Fixed sync cycle management during cycle purge.
- Improved notifications for index saves and archive jobs. They were wronlgy reported as regular backup jobs.
- The CLI [ArchiveSelection addfrom] now properly honours incremental setup.
- Added new CLI [ArchiveSelection addfile] and [ArchiveSelection addfileabs] commands.
- Recompiled Linux Sqlite package in 32bit to allow 64bit file offsets.
- Fixed cross-platform character conversion introduced in V5.2.1 by changes for special archive restores.
- Fixed B2Go template propagation to local workstation.
- Fixed large-file support for V4-compatibility indexes.
- Fixed FinderInfo issues for restores on Helios volumes when operating in 64bit mode.
- Fixed B2Go automatic update when several workstations start updating the same time (overlap).
Version 5.2.1 released. Changes since last release:
- Fixed remote update of older (i386) Mac OSX clients.
- Fixed handling of volumes to backup containing blanks in their mount points.
- Ignore previous configuration of the user-defined mail script; always use our own mail sender(s).
- Improved browsing for archive indexes to exclude files located on unknown (recycled) volumes.
- Fixed browsing P4 backup index.
- Fixed changing index option in P4 backup index.
- Fixed file version restore in P4 backup index.
- Fixed timing problem with Firefox browser within restore option window.
- Fixed Windows B2Go client login problem.
- Fixed setting old volume to suspect after automatically recycling it during a running job.
- Improved Windows installation procedure.
- Improved volume import to allow continuation of the import in case of job error or cancellation.
- Volume mode can now be manually re/set to Migrated.
- Added compatibility for CLI restores to allow restoring files at the original (physical) path if no relocation was requested, as for version 5.1 and below.
- Added log traces for per-job activation and completion callbacks.
- Fixed restore relocation for archive jobs on Windows.
- Fixed default hint for relocation path for archive restores in GUI.
- Fixed id/path translation for remote archive restores.
- Removed UUID (GUID) handling for ZFS filesystems since it resulted in occasional duplicates thus yielding two or more different file systems having the same UUID.
- Fixed occasional crashes when restoring from archive.
Version 5.2 released.
For new features, please refer to the 5.2 Features List.