Roadmap

v0.21.3

Due in 7 days (09/15/2010)

29%

1 closed (14%)   6 open (86%)

Related issues
Bug #295: mds: can't rmdir due to dir size underflow
Bug #326: OSD crash PG::IndexedLog::unindex
Bug #329: mds: mislinked dentry found during journal replay
Bug #344: cfuse should pass all qa tests
Bug #374: mon: osd will null addr added to map
Bug #383: cfuse: reimplement readdir
Bug #386: mon: class library out of whack

v0.22

Due in 23 days (10/01/2010)

stable clustered mds

49%

21 closed (47%)   24 open (53%)

Related issues
Bug #29: mds: rfiles underflow
Bug #30: multimds: slave_request on getattr
Bug #35: osd: pg split should queue transaction(s) under appropriate sequencer(s)
Bug #55: osd: fix transition from snaps -> no snaps -> snaps
Bug #213: non-idempotent transactions (clone) under ext3 may not replay correct result
Bug #247: osd: allow separate interface for internal cluster traffic
Bug #262: osd: make sync work on ext4
Bug #316: mds: track dirty snapped metadata without faking a cap
Bug #318: Multi-MDS snaptest-2 fails
Bug #319: allows snaps in root directory
Bug #325: writes starve reads
Bug #346: back-off on clock drift errors
Bug #351: mds: fix cow on xattrs
Bug #355: snapshots fail to maintain data integrity
Bug #356: mds: cap writeback/release race with cow_inode
Bug #360: mds: head/snapped snap_cap linkage may cross mdss
Bug #362: mds: rejoin crashes on snaptest-2 workload
Bug #370: mds: need to re-eval caps after import
Bug #372: 2-Monitor election fight
Bug #373: Memory leak in cfuse
Bug #393: mds: migration vs locking
Bug #394: mds: scatterlock scatter dir accounted_fragstat writeback
Bug #395: mds: interval_set assert(0) during journal replay
Bug #399: mon: trim laggy standby mdss from map
Feature #82: mon: osd failure smarts
Feature #97: qa: better snap tests
Feature #138: Try out tcmalloc
Feature #185: mds: set file layout policy on directory hierarchy
Feature #195: monitor naming and removal support
Feature #235: cfuse: allow specifying server path to mount on command line
Feature #238: cfuse: log to file if run as daemon
Feature #245: Logging to syslog
Feature #286: librados: set read chunk size
Feature #298: osd: "watch" operation
Feature #303: objecter: limit in-flight ops and/or bytes written
Feature #321: allow dns names for monitor switch
Feature #352: unit tests to check snap parents
Feature #359: osd: use new btrfs snapshot ioctls
Feature #364: osd: leave gap after journal entries to avoid full disk rotation
Cleanup #124: msgr: change protocol handshake to exchange in_seq
Cleanup #236: cfuse: clean up mds session handling
Cleanup #311: osd: remove read(len=0) full object behavior
Cleanup #324: standalone ceph-client-tools package
Tasks #365: test snaptests against single mds failure
Tasks #366: test snaptests against clustered mds failures

v0.23

Due in 54 days (11/01/2010)

0%

0 closed (0%)   1 open (100%)

Related issues
Bug #354: Detect errors during transactions

v1.0

0%

0 closed (0%)   13 open (100%)

Related issues
Bug #250: mon: delete old states to avoid filling disk
Bug #369: crush: calculate separate parent mappings for all possible hierarchies
Feature #20: mds: allow client reconnect while up:active
Feature #86: mds: fsck
Feature #88: mds: change stray commit strategy to avoid rolling stray dir commits
Feature #89: mds: put inode dirty fields in dirty_bits_t to reduce memory footprint
Feature #90: mds: don't sync log on every clientreplay request
Feature #94: mon: monitor some osds
Feature #95: mon: adjust overload based on osd disk utilization
Feature #314: osd disaster recovery tools
Feature #315: mds disaster recovery tools
Feature #340: large directories, directory fragmenting
Feature #358: Revert to snapshot

Linux kernel client - v2.6.36

45%

8 closed (42%)   11 open (58%)

Related issues
Linux kernel client - Bug #28: gracefully fail on fill_trace errors
Linux kernel client - Bug #69: ceph: ffff88001976ba50 auth cap (null) not mds0 ???
Linux kernel client - Bug #111: handle EAGAIN from osd
Linux kernel client - Bug #118: clean pages when throwing out dirty metadata on session teardown
Linux kernel client - Bug #119: avoid looping connect/retry errors on console
Linux kernel client - Bug #147: lockdep: possible irq lock inversion dependency w/ osdc->request_mutex and con->mutex
Linux kernel client - Bug #216: splice_dentry error
Linux kernel client - Bug #217: client needs to open sessions to mds export targets
Linux kernel client - Bug #283: ceph_add_cap: couldn't find snap realm, NULL ptr deref
Linux kernel client - Bug #304: GPF in writepages_finish
Linux kernel client - Bug #333: dup flushsnap messages
Linux kernel client - Bug #335: xattrs not coherent wrt snaps
Linux kernel client - Bug #384: crash in splice_dentry
Linux kernel client - Bug #392: writes beyond 4GB wrap on 32 bit clients
Linux kernel client - Feature #19: rbd
Linux kernel client - Feature #23: fcntl/flock advisory lock support
Linux kernel client - Feature #24: mdsc: preallocate reply msgs
Linux kernel client - Feature #25: mdsc: mempool for cap writeback?
Linux kernel client - Cleanup #233: replace mount option fsid{major,minor} with fsid=<string>

Linux kernel client - v2.6.37

0%

0 closed (0%)   4 open (100%)

Related issues
Linux kernel client - Bug #206: make a 'soft' mode
Linux kernel client - Bug #227: caps flush races with caps export
Linux kernel client - Feature #294: honor PAUSE* flags in osdmap
Linux kernel client - Feature #387: expose directory subtree partition/replication/fragmentation via ceph.* xattrs