Tags: tojoyccnu/moosefs
Tags
MooseFS 3.0.115 * (cs) fixed chunk locking issues in chunkserver (lock timeout, missing release etc.) * (all) changed to hex all chunkid logged to syslog (only in changelog it is left in decimal format) * (cs) changed port number from hex to dec in replicator module * (mount) added printing open/create mode to the oplog * (mount) fixed opening file after using lookup cache * (master) added ignoring empty filenames
MooseFS 3.0.114 * (mount) fixed trunacting files open with O_APPEND flag (bug itntr. in 3.0.113, issue moosefs#368) * (cs) added disk rebalance when all designated source disks are already empty (issue moosefs#364) * (mount) added additional info to '.params' file (versions of mfs,libfuse itp.) * (master+cs) added check against EACCES after lockf (according to Linux man - such stupid error can be returned here - issue moosefs#369) * (master) fixed access checks in snapshot * (cgi+cli) removed using supervisor (only available in PRO version) * (cs) fixed deadlock condition on mutexes 'folderlock' and 'hashlock' (exists since 3.0.110) * (tools) fixed oveflow protection in number parsing function * (cs) fixed reporting to master duplicates with newer version * (cs) changed job queue policy (some tasks are not limited now) * (master) fixed reaction to status NOTDONE received from chunkservers * (master) fixed syncing and closing negative file descriptor in bgsaver * (mount+master) fixed handling keepcache and direct flags (related to issue moosefs#374) * (cgi) changed time format (issue moosefs#197) * (cs) added logging info when '.chunkdb' is not written to disk * (master) fixed alphabetical order of commands detected in changelog * (cs) changed queue limit to max workers (limited dynamically) * (cs) fixed automatic chunkserver removal in master * (master) added meta version increment in chunks_set_version * (cs) added error detection during writing '.chunkdb' * (all) added build id to 'what' strings * (nbd+cgiserv) added chdir("/")
MooseFS 3.0.113 * (master) changed condition that decides if master should wait for more chunks during I/O * (mount) delayed setting channel for fuse notifications (very rare segfault in libfuse2 during init) * (client+master) changed conditions for ancestor test in getattr (issue moosefs#350) * (cs) added parameter for number of chunks to be send in single register packet * (cs) fixed reporting damaged chunks in testing function (issue moosefs#352) * (client) changed open test in setattr (related to issue moosefs#350) * (master) fixed handling truncate for open files (related to issue moosefs#350) * (master) added uid mapping in setfacl function * (master+client) added support for atomic truncate with open * (master+client) fixed keep cache conditions * (cs) introduced official label format defined in chunkserver configuration * (nbd) added readonly mode and locking * (cs) fixed condition in choosing disks for internal rebalance * (mount) added workaround in access for a bug in FreeBSD kernel (issue moosefs#354) * (cs) fixed master reconnection conditions (reload usually shouldn't cause reconnection) * (master) changed disk removal detection algorithm (issue moosefs#356) * (cs) fixed calculating size limits * (cs) added handling inode limits in local filesystems on chunkservers (issue moosefs#358) * (master) added topology grouping when new chunks are about to be created (prefer closer servers) * (master) added more error messages to bgsaver * (cs) added changing subfolder during internal rebalance (related to issue moosefs#326) * (master+cs) added optional logging of long function execution times * (mount) turn off dentry invalidator for Linux kernels >= 4.19 (related to issue moosefs#357) * (master) changed condition that checks timestamp in changelogs (less prone to small clock differences) * (cgi+cli) changed mark for removal state name UNKNOWN->PENDING (related to issue moosefs#359) * (client) decreased max IDLE time in writer module from 1s to 0.1s * (client) update inode in dentry invalidator * (tools) fixed mfscopyeattr,mfsseteattr and mfs*arch tools * (master+client) added support for new eattrs: immutable,appendonly,undeletable * (client) silenced "kern.proc.filedesc" syslog messages on FreeBSD (issue moosefs#360)
MooseFS 3.0.112 * (cs) silence stupid compiler warning * (client) fixed handling LOCKED and EAGAIN status in readdata * (client) added handle info to oplog messages * (all) added support for disabling individual filesystem commands in exports * (client) added handling read/write/readdir disables on client side (better error messages) * (client) added session paramaters to '.params' file * (tools) fixed packet reallocation error * (master) changed syslog message for locked chunks * (client) added dentry invalidator (needed in Linux with kernel < 4.19 - EBUSY bug) * (master) fixed memory leaks in xattr and posixacl modules * (cs) when chunk can't be located always send to master info about lost chunk * (supervisor,master,cs) fixed buffer overrun in mastersupervisor code (HA only, intr. in 3.0.107) * (master) changed algorithm of reusing csid in chunk module (adding released csid to the end of free list) * (all) added support for STRICT and LOOSE mode in KEEP and ARCHIVE * (master) fixed segfault during appending chunks of file with positive length and no chunks * (metadump) fixed bugs introduced in verson 3.0.106 * (master) added protection between neverending desyncing between newer LEADER and older FOLLOWER (HA only) * (master) fixed slices with 'to' set to 0 in mfsappendchunks * (master) removed starting protection time from client communication module * (man) fixed typo in mfsappendchunks man page
MooseFS 3.0.111 * (cs) added protections against segfaults (issue moosefs#342) * (cs) removed unnecessary error messages
MooseFS 3.0.110 * (cs) fixed bug that may lead to creating much more chunks in one subfolder (issues moosefs#319, moosefs#326) * (cs) added option for controlling subdirectory selection algorithm for newly created chunks (issues moosefs#319, moosefs#326) * (cs) prevent potential dumping deleted chunks to 'chunk.db' * (client) added options for setting behaviour when master returns unavailable chunks or no space status (issue moosefs#322) * (cgi) fixed logo and icon (file type for favicon.ico, change to current logotype – issue moosefs#328) * (master) turning off sending NOPs to unregistered clients (matoml module) * (cs) fixed reporting wrong number of hours in duplications removal message * (cs) redesigned duplicates removal procedure to do it in groups of 100 files (issue moosefs#327) * (cs) introduce mypread and mypwrite macro (code cleaning) * (cs) added check for changed crc before every chunk file close (just in case) * (cs) never use open chunk as a chunk duplicate that needs to be deleted * (cs) never use disk with pending duplicates as destination disk during internal rebalace (issue moosefs#327) * (cs) fixed condidtion in internal rebalance procedure (issue moosefs#327) * (cs) added option for setting duplicates retention time in hours (issue moosefs#333) * (cs) fixed memory leak in removing disk with duplicates * (cs) removed unnecessary attribute reading when chunk duplicates are detected (issue moosefs#327) * (cs) changed allocation of duplicated chunks from malloc to mmap, decreased memory usage by duplicates * (client) reading in O_WRONLY no longer causes EACCES (might be related to issue moosefs#334) * (metadump) printing dots for non-printable characters made optional (issue moosefs#332)
MooseFS 3.0.107 * (master) added NOP command to changelog * (master+supervisor) added time check * (cs) fixed behaviour when job queue is full * (cli+cgi) added time data to master servers * (cs) fixed NOP packets ping-pong * (client) fixed NOP flooding in write module * (cs) fixed race condition in bgjobs module (jobhash) * (cs+mount) added new status returned to mount when queue is full and support for it in client module * (mount) added initial support for libfuse3 (related to issue moosefs#302) * (mount) added support for readdirplus (related to issue moosefs#302) * (mount) workaround for bug in fuse (wrong lock_owner in flush - issue moosefs#305) * (mount) fixed deadlock in readdata.c (closes issue moosefs#306) * (mount+master) fixed few small issues related to POSIX compliance (checked by pjd) * (all) changed prefered C standard from C99 to C11 and added tests for atomics * (mount) move read/write structures initialization from open to read/write * (mount,mfsio) added new inode->fleng map * (configure) added option disable_all, added summary at the end * (freebsd) detecting new fuse version in the kernel and adjusting behaviour * (tools) better handling of unrecognized options * (master) better hard links handling * (master) added limit for number of hardlinks allowed for one object * (master) added option for automatic removal of unused chunkservers (issue moosefs#247) * (metadirinfo) fixed segfault when output filename is not specified * (chunktool) added file name fixing option (issue moosefs#276) * (all) changed chunk replication and deletion charts into two color charts (successful/unsuccessful - issue moosefs#273) * (all) added new charts with number of chunk creations and internal changes (also successful and unsuccesful) * (master) added ignoring chunk with id zero sent from chunkserver * (metadump) added option for dumping parts of metadata * (master) added more logs to background data write subprocess (connected with issue moosefs#280) * (master) fixed LEADER/FOLLOWER desync on SETCHUNK command (pro only) * (cli+cgi) fixed shebang in python scripts (preference for python3 in configure) * (master) changed pipe in bgsaver to nonblocking mode * (master) added check if the background saving process is alive * (master) added lockfile for background saving process * (master) added protection against time going back in the free inodes list * (master) added protection in FOLLOWER against not keeping up with the LEADER (pro only) * (master) changed condition for removing chunk from data structures (all chunkservers have to be connected) * (master) added 'restore' run mode (works similar to 'mfsmaster -a' but without running daemon) * (master) take write replication counters into account in server ordering for new chunk creations * (master) changed calcualting replication limits in chunk rebalance procedure (using actual usage difference)
MooseFS 3.0.105 * (nbd) fixed segfault when filename not specified * (client) fixed size of waiting threads counter (very rare deadlock in write module) * (cli) fixed status returned in case of master name that can't be resolved (issue moosefs#238) * (nbd) added option for specifying subfolder on MFS for block devices (issue moosefs#252) * (netdump) fixed using deprecated function pcap_lookupdev * (cli) fixed 'free space' in plain mode
PreviousNext