What’s New in StorNext 6.3.0
The StorNext 6.3.0 release provides new features and enhancements listed in the section New Features and Enhancements in StorNext 6.3.0 and also provides software fixes listed in the section Fixed Issues and Enhancements Addressed in StorNext 6.3.0.
Support Long Object IDs on Object Import
Prior to StorNext 6.3, objects imported with the command fsobjimport were restricted to those with a key length that did not exceed 64 bytes. Beginning with StorNext 6.3, we enhanced the tertiary storage manager (TSM) infrastructure to support object keys with a larger length, up to 1,024 bytes.
Read Stream Reservation Capability
Beginning with StorNext 6.3, you can reserve a subset of the object storage streams for file retrieves.
Use the StorNext GUI to reserve streams for reads when you configure an object storage controller. You can configure the read streams in the Controller section of the Configuration > Storage Destinations > Object Storage menu. See Configure Object Storage and Cloud Destinations.
You can also use thecommand line interface (CLI) to run the fsobjcfg command with the -s option to configure a read stream count. You can exceed the number of streams used for retrieves, but you cannot use the reserved streams for non-retrieve requests.
See the fsobjcfg command in the StorNext 6 Man Pages Reference Guide.
Enhancements to the fsCapacityThreshold Configuration File Parameter
Beginning with StorNext 6.3, StorNext automatically monitors the used capacity of the HA shared file system and generates a RAS event when the file system reaches a capacity of 85% or higher.
Note: StorNext continues to generate a Severity 2 RAS event once an hour until the used capacity level falls below 85%.
The percent at which a RAS event is generated is controlled by the HA shared file system’s configuration file parameter fsCapacityThreshold.
Note: If the fsCapacityThreshold parameter is not present in the configuration file or if it is set to zero (0), then the default of 85% is used.
See How to Configure the fsCapacityThreshold Configuration File Parameter for additional details.
New Extended Attributes
Beginning with StorNext 6.3, a series of new extended attributes have been added, as follows:
|user.com.quantum.retrieves||This attribute can be read or set. It determines whether a client has the ability to retrieve files or not. To configure, you must write the value enabled (allow retrieves) or disabled (deny retrieves).|
|user.com.quantum.snsm||This attribute can be read. It provides the current state of a managed file and also provides different descriptions:
|com.quantum.rpl||This attribute can be read. It provides a reverse path lookup of a given file.|
|user.com.quantum.location||This attribute can be read. It provides a json response that includes the media type backing a managed file and also provides the media UUID of the media the file is written to when it is truncated.If the file is an object store file, the object store path is given. If an affinity is assigned to the media of the file, the affinity is included in the json response.|
|user.com.quantum.offline||This attribute can be read. It returns a true or false response to whether a file is offline.|
|user.com.quantum.clients[.classname]||This attribute can be read. It returns a json response of client information that have been mounted with a specific mount option.|
|user.com.quantum.sn_rest_uri||This attribute can be read. It returns the URI for the file system manager’s (FSM’s) rest interface.|
|user.com.quantum.sn_rest_challenge||This attribute can be read. It returns the challenge string for the FSM’s rest interface.|
|user.com.quantum.blocklocations||This attribute can be read. It returns the stripe group number of a file’s location.|
|user.com.quantum.affinity||This attribute can be read. It returns the affinity associated with a file.|
On systems running Linux, you can run the command getfattr to access or write to extended attributes.
Note: You must enable namedstreams on the file system and you must also remove the user namespace.
tool for Windows.
|CvApi_GetXattrByName(int fd, XattrRequest_t *xreq)||This API takes a file descriptor and a XattrRequest_t pointer. The XattrRequest_t structure consists of two buffers:
The xattr_name is confined to 117 bytes and the xattr_value is confined to 4,096 bytes.The xattr_value should not have any data written to its buffer.By calling this function, the xattr_value buffer is filled and returned to the user.
|CvApi_SetXattrByName(int fd, XattrRequest_t *xreq)||This API takes a file descriptor and a XattrRequest_t pointer. The XattrRequest_t structure consists of two buffers:
The xattr_name is confined to 117 bytes and xattr_value is confined to 4,096 bytes.This function only works with xattrs that are able to be written to.Both buffers must have data in order to properly use this function.The structure is returned back to the user with the same data it was supplied with.
To build a program with Quantum’s external API, you must include extapi.h and cvapi.h in the source. You must also include cvextapi.lib when you build the program for Windows.
By using these functions, a Windows user can emulate the use of extended attributes without them being an interposed component of the file system and Windows.
Note: Not all of the extended attributes are available for use through Windows and not all of the attributes are able to be read or written depending on the user requesting and the settings on the file.
|Operating System||Change Request Number||Service Request Number||Description|
|All||32784||474686,1223902, 1197170, 1286816, 1559740, 3545946, 3642968||Excessive tac log messages – no media found to meet user criteria|
|All||62966||3681170, 344347, 478334, 480193||MSM can wrongly identify the running process, leading to system crash|
|All||65786||503315||Update TSM infrastructure to support long/alternate object names|
|All||68878||n/a||tac log shows mysql connection error when querying for Qcloud configuration|
|All||70955||399606 489960||FSM OpHangLimitSecs exceeded VOP-Class-0 Type-4 Subtype-11RevokeWait -> set_wait_racing|
|All||71103||402207, 504661, 506855||Need to monitor used capacity of HAshared file system more closely|
|All||73252||n/a||provide streams reservation capability for object get operation|
|All||73295||424025||Security Scanner cause FSM on secondary MDC to die Weekly|
|All||73322||463088||HA shared FS expansion via Stripe Group Actions leads to non-recommended configuration, when HA shared FS has a mixed metadata and data SG and no exclusive MD SG|
|All||73674||478052||Name service attempts to send udp messages to port zero on startup|
|All||73676||501660||ANTF & LTFS tape fsexport failure after database error|
|All||73896||n/a||GUI: Support for read stream reservation count in controller configuration|
|All||74108||481323||Found several minor issues in wsar_client with the -s option, -l option, and documentation|
|All||74111||482099||attempting to start second instance of snstatd from command line leads to qustat failures|
|All||74139||474784||fsm memory bloating: filesystem hit OOM upon starting servicing change_tree REST call from flexsync|
|All||74164||482099, 502217||fs_resourced locks up and requires restart of TSM|
|All||74169||477772||fsmedcopy/fsfilecopy from tape to S3-compatible medium fails and marks it as write-protect|
|All||74174||484016, 494543, 504800||StorNext should warn when metadata space is low|
|All||74175||484016||cvupdatefs failed to add a new metadata stripe group because there were 0 free inodes|
|All||74228||495024||‘/usr/cvfs/lib/snnas_control start’ does not start snnas_service|
|All||74231||479813||syncha can delete all local config files if the shared filesystem is dismounted while it is running|
|All||74290||488616, 493109||NAS beyond 2.2.0 and SNFS work — Windows Directory notifications (samba <-> SNFS) not delivered.|
|All||74453||490072||fs_feature core dump due to segmentation fault when running archive_cmp|
|All||74476||480485||Non-existent or truncated display of existing file system quotas in the snquota CLI and StorNext GUI|
|All||74502||n/a||Add FS_CLUSTER_LIMIT_XXX to fs_syparm.README file|
|All||74509||491986||syncha too noisy: reports “unable to determine hostdatadir for peer” in single mode|
|All||74559||490052||Incorrect name server info messages when using fsforeignservers|
|All||74609||477352||A relocate policy will not clean all candidates that it should|
|All||74618||487263||D2D Relocation – Nightly class based relocation does not work if tiers are named out of order|
|All||74623||490052||StorNext should alert Customer if current mounted file system volumes are out of date and need to be rescaned|
|All||74629||493309||fsmpm: PANIC: fsmpm ASSERT failed “ntohs(pmsg->nsm_data16) & NSS_HB_EXTENDED_OK”|
|All||74632||487263||Fix relocation-by-class policy so that it doesn’t overwrite its policy start time|
|All||74633||487263||Fix fs_tierman so that it honors the configured policy class Reloc Min Time when it is a value less than 1 day|
|All||74644||496096||recover_xattr_payloads() does not honor snbt_lookup() API causing snbt_payload_expect_dupes():ASSERT(FALSE)|
|All||74645||494879||kernel panic w/ kernel BUG at fs/dcache.c:667, exception in shrink_dcache_for_umount_subtree() during cvfs umount|
|All||74646||492529||wsar_agent segfault occurred due to unhandled null pointer exception in wsardb_CleanJobs()|
|All||74652||n/a||FS_BLOCKING_READS setting not working|
|All||74665||498164||immutable bit on root prevented starting file system because qbm create/remove of directory|
|All||74796||501350, 474104||snaudit “Failed to read tablet cluster” file /scm/nightly/VM-RedHat7-x86-64-SP3-0/sn/snfs/qrtree/qrtree_backend_qrdb.c, line 1058|
|All||74808||493941||Reading data from Media with large clusters can lead to issues finding the BOF and report Invalid label data|
|All||74822||498657||After upgrade from 6.1 to 6.2 ala report main: Database error during insert: code: 1406, message on insert into sl_admin_log|
|All||74842||502251||GUI: deprecated media types in MSM database can cause vault create failure in UI after upgrading to 6.2.0|
|All||74869||493530||Remote DDM processes fail and generate cores after upgrade to StorNext 6.2.0|
|All||74871||502252, 497249, 503172, 498392||CPU or system lockup when kernel oplocks enabled in NAS|
|All||74873||503173||fsm segmentation fault upon sgmanage resize, due to unhandled exception in move_remainder_space()/inode_btree_start_slice()|
|All||74889||499858||StorNext close locking causes nfs server to consume millions of file entries|
|All||74890||503808||fsm panic “OpHangLimitSecs exceeded” whilst vacating stripe group in progress, due to deadlock between free_pending_inode and vacate_stripegroup threads|
|All||74891||505693||sgmanage displays the wrong size for several counters when the fs block size is not 4k|
|All||74892||503808||fsm panic upon ASSERT failed “sl < pgroup->pg_slices” snfs/fsm/alloc.c, line 3058, after a stripe group offload/vacate was interrupted|
|All||74902||n/a||TSM checkMediaAvailabilityTsm health check script must detect non-tape store availability|
|All||74915||488583||dm_info attributes not set correctly after a file move. endtime of old filecomp not reset|
|All||74948||503993||snrestore fails due to duplicate entries (but with different media) in the snbackup_manifest file|
|All||74952||n/a||The new scan functionality being used in fsaddrelation and rebuild policy can hang.|
|All||74986||505693||sgadd miscalculates stripe breadth when adding a stripe group to a file system with an original block size other than 4k|
|All||75032||501660||GUI : Export Files job report incorrect : job overall status is reported as success, although errors were encountered|
|Linux||73452||460749, 465920||Stopping managed FSM hangs Stornext GUI|
|Linux||74545||501127, 493912, 501127||Ubuntu Kernel 220.127.116.11 new get_user_pages() signature, cvfsbuild fails error: too many arguments to function ‘get_user_pages’|
|Linux||74687||497732||Support for CentOS/RHEL 7.6|
|Linux||74809||501686, 507321||snprobe not returning with options -cdlqsC -h localhostCaution: StorNext 6.2 included new cloud-based monitoring tools to simplify and streamline system management when combined with Web Services. In certain situations, Cloud Based Analytics (CBA) reporting generates repetitive snprobe returns that never complete, resulting in system memory being depleted. If this occurs, Xcellis systems might unexpectedly reboot. See StorNext Product Bulletin 105 for additional details.|
|Windows||46895||3353988||SNFS latency-test give extremely bad performance on Windows node.|
|Windows||72817||454915||Stornext Windows GUI need to allow the creation of fsforeignservers|
|Windows||73687||455957 505429||IsLocalSid() Requires that Administrator account to exist.|
|Windows||74181||479521, 487047||max length russian characters|
|Windows||74583||480485||Photoshop deletes the original file if Quota is exceeded – FileSetAttr() problem|
File System Server: Server Revision 6.3.0 Build 84384 Branch branches_6.3.0 BuildId E Built for Linux 3.10.0-514.el7.x86_64 x86_64 Created on Thu Aug 1 16:18:24 MDT 2019 Built in /scm/nightly/VM-RedHat7-x86-64-SP3-0/sn/buildinfo File System Client: Client Revision 6.3.0 Build 84384 Branch branches_6.3.0 BuildId E Built for Linux 3.10.0-514.el7.x86_64 x86_64 Created on Thu Aug 1 16:18:24 MDT 2019 Built in /scm/nightly/VM-RedHat7-x86-64-SP3-0/sn/buildinfo