summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* mon: fix allocation of low global_ids after mon restarthistoric/aleung_mds_securitySage Weil2009-12-022-7/+10
| | | | | We need to set last_allocated in init(). Also avoid the -1 casting and just use 0 instead.
* test_ioctls: print preferred_osd value from GET_LAYOUT ioctlSage Weil2009-12-022-2/+3
|
* hash: fix whitespaceSage Weil2009-12-021-4/+4
|
* mds: set mdr->in[n] in rdlock_path_xlock_dentrySage Weil2009-12-021-0/+1
| | | | | | | | | | | | | | | This keeps things consistent with what we'd get with path_traverse. Namely, it means that if done_locking gets set to true, then a future call to rdlock_path_pin_ref can bail out and return mdr->in[n] and get a correct value. This fixes touch mnt/a ls mnt touch mnt/.ceph mds hang on the create request.
* mds: do not include /.ceph in readdir resultsSage Weil2009-12-022-0/+4
|
* librados: set id in messengerGreg Farnum2009-12-011-1/+1
|
* rados: read benchmark is asynchronous.Greg Farnum2009-12-011-32/+157
|
* Change hump-backed variable names.Greg Farnum2009-12-011-19/+21
| | | | Initialize data->in_flight!
* rados: pull sequential benchmark into its own functionGreg Farnum2009-12-011-41/+56
|
* rados: remove iTime from object names for complete determinismGreg Farnum2009-12-011-6/+3
|
* rados: Handle errors slightly betterGreg Farnum2009-12-012-13/+30
|
* rados: split up function into more subfunctionsGreg Farnum2009-12-011-142/+159
|
* rados: Deterministically fill data, reverse count:time orderGreg Farnum2009-12-011-3/+8
|
* librados: minor commentsGreg Farnum2009-12-011-0/+3
|
* rados: Move variables into data structure; rename to be genericGreg Farnum2009-12-011-88/+85
|
* rados: Move benchmark functions into separate fileGreg Farnum2009-12-012-269/+291
|
* qa: Rename a file for future clarity.Greg Farnum2009-12-011-0/+0
|
* qa: specify test path as argument ot runallone.shSage Weil2009-12-012-5/+5
|
* config: preserve argv[0] if appropriateSage Weil2009-12-011-1/+4
| | | | This makes 'cfuse' show up in 'df' output.
* qa: test mdstable failuresSage Weil2009-12-013-10/+27
|
* todoSage Weil2009-12-011-173/+3
| | | | the lockdep was probably due to the leak of request_mutex?
* mds: inject mds failure during mdstable operationsSage Weil2009-12-014-0/+23
|
* mds: log successful client reconnectSage Weil2009-12-011-4/+8
|
* sepia: asdfSage Weil2009-12-011-2/+2
|
* mds: journal/replay rollback on MDSTableServerSage Weil2009-12-013-0/+9
| | | | We journal prepare and commit; need to do rollback to!
* mds: track _note_prepared/committed in normal path, tooSage Weil2009-12-012-1/+5
| | | | This was broken by cd6f3817.
* mds: pass do_eval through to adjsut_subtreeauthSage Weil2009-12-011-1/+1
|
* vstart: add [mds.a] section when no -xSage Weil2009-12-011-2/+6
|
* mds: drop requests on closed|closing|etc sessionsSage Weil2009-12-011-1/+7
|
* ceph.spec: mostly in line with .debSage Weil2009-11-251-61/+28
| | | | No packages for radosgw or libcrush/ceph/rados, though.
* object: add missing includeSage Weil2009-11-251-0/+1
| | | | For sprintf().
* mds: ignore reconnect if session state is closedSage Weil2009-11-251-1/+1
| | | | | On reconnect, if we didn't have a session from the journal, our Session state will be 'closed'. Act accordingly.
* todoSage Weil2009-11-241-0/+112
|
* mds: don't renew caps (or touch_session) on closed sessionSage Weil2009-11-242-8/+16
| | | | Also assert that we're on the state list before touching.
* mds: less --, more -Sage Weil2009-11-241-2/+2
|
* mds: update pending_for_mds during table server replaySage Weil2009-11-243-4/+12
| | | | | Otherwise we our replay of AGREE after recovery is wrong; we just replay what was saved, without taking additional journaled items into account.
* mon: add 'mds set_state <gid> <stateid>'Sage Weil2009-11-241-0/+14
| | | | For mdsmon disaster recovery :)
* mon: avoid mds state update when clearing laggy flagSage Weil2009-11-241-29/+25
| | | | | | Do not clear laggy flag (or otherwise get into prepare_update) if the mds hasn't seen the latest map. Previously we could go to clear laggy and also revert to an old mds state.
* vstart.sh: fix -k for mds loadingYehuda Sadeh2009-11-231-4/+2
|
* vstart.sh: fix -k (cmds init)Yehuda Sadeh2009-11-231-2/+2
|
* msgr: fix bad unlock on tcp_write errorSage Weil2009-11-231-4/+2
|
* mds: fix replay of hard linksSage Weil2009-11-231-1/+1
| | | | | Use dir->link_remote_inode, not raw function, so that the CDir accounting gets set up.
* mds: clear session before remove_sessionSage Weil2009-11-201-2/+1
| | | | remove_session may drop the last ref.
* mds: do not eval subtree root during rename_prepareSage Weil2009-11-203-11/+12
| | | | | | | | | | | The eval may lead to a scatter_writebehind, which screws up the journal ordering. Since the inode is also locked, we will end up evaling it later anyway, so don't worry about it. I'm a bit unsure about the adjust_subtree_after_rename... that is during the rename_apply stage (after the log entry committed), so i think it's safe to do a scatter_writebehind here.. since we're not between a predirty and submit_entry.
* todoSage Weil2009-11-201-54/+73
|
* mds: fix client purge crashSage Weil2009-11-201-11/+1
| | | | | | | | | | | | | | | | | | | | | | set_state moves the session to the per-session lists; do this _before_ remove_session (which removes the session entirely). Otherwise we fail an assert when deleting the session: mds/SessionMap.h: In function 'virtual Session::~Session()': mds/SessionMap.h:141: FAILED assert(!session_list_item.is_on_xlist()) 1: ./cmds(_Z18__ceph_assert_failPKcS0_iS0_+0x36) [0x97fc87] 2: ./cmds(_ZN7SessionD0Ev+0x47) [0x73d817] 3: ./cmds(_ZN16RefCountedObject3putEv+0xdb) [0x712c3b] 4: ./cmds(_ZN10ConnectionD0Ev+0x3a) [0x713f88] 5: ./cmds(_ZN10Connection3putEv+0x3d) [0x71294b] 6: ./cmds(_ZN15SimpleMessenger4PipeD1Ev+0x7f) [0x7130e1] 7: ./cmds(_ZN15SimpleMessenger6reaperEv+0x228) [0x7048ba] 8: ./cmds(_ZN15SimpleMessenger4waitEv+0x28) [0x7059dc] 9: ./cmds(main+0x305) [0x6f9345] 10: /lib/libc.so.6(__libc_start_main+0xe6) [0x7f2091e715c6] 11: ./cmds [0x6f8e39] NOTE: a copy of the executable, or `objdump -rdS <executable>` is needed to interpret this.
* mon: fix use-after-free in mds mon blacklistingSage Weil2009-11-201-1/+1
|
* mds: warn if replay used ino doesn't match session's nextSage Weil2009-11-202-1/+12
|
* mds: fix reconnect raceSage Weil2009-11-203-2/+12
| | | | | | | Don't ignore client replay or session msgs if we are moving to reconnect state. The client may get the mdsmap before us and send things our way, so take them if we are reconnect, or if we want reconnect.
* mds: remove dead session request list trim waiter codeSage Weil2009-11-201-14/+0
| | | | I don't even remember what this was originally for.
View Lorry Controller Status