1 2004-03-06 multipath-tools-0.1.1
2 * include dlist.h in multipath main.c (PM Hahn)
3 * typo in hotplug script (PM Hahn)
4 * pass -9 opt to gzip for manpages (PM Hahn)
5 2004-03-05 multipath-tools-0.1.0
6 * add the group_by_tur policy
7 * add the multipathd daemon for pathchecking & DM hot-reconfig
8 * multipath doesn't run twice
9 * massive cleanups, and code restructuring
10 * Avoid Kernel Bug when passing too small a buffer in do_inq()
11 * Sync with 2.6.3-udm4 target synthax (no more PG prio)
12 2004-02-21 multipath-018
13 * From the Debian SID inclusion review (Philipp Matthias Hahn)
14 * use DESTDIR install prefix in the Makefile
15 * add man pages for devmap_name & multipath
16 * correct libsysfs.h includes
17 * fork the hotplug script in its own shell
18 * Sync with the kernel device mapper code as of 2.6.3-udm3
19 ie. Remove the test interval parameter and its uses
20 * Remove superfluous scsi parameter passed from hotplug
21 * Add the man pages to the [un]install targets
22 2004-02-17 multipath-017
23 * remove the restrictive -f flag.
24 Introduce a more generic "-m iopolicy" one.
25 * remove useless "int with_sysfs" in env struct
26 2004-02-04 multipath-016
27 * add a GROUP_BY_SERIAL flag. This should be useful for
28 controlers that activate they spare paths on simple IO
29 submition with a penalty. The StorageWorks HW defaults to
30 this mode, even if the MULTIBUS mode is OK.
31 * remove unused sg_err.c
32 * big restructuring : split devinfo.c from main.c. Export :
33 * void basename (char *, char *);
34 * int get_serial (int, char *);
35 * int get_lun_strings (char *, char *, char *, char *);
36 * int get_evpd_wwid(char *, char *);
37 * long get_disk_size (char *);
38 * stop passing struct env as param
39 * add devmap_name proggy for udev to name devmaps as per their
40 internal DM name and not only by their sysfs enum name (dm-*)
41 The corresponding udev.rules line is :
42 KERNEL="dm-[0-9]*", PROGRAM="/sbin/devmap_name %M %m", NAME="%k", SYMLINK="%c"
43 * remove make_dm_node fn & call. Rely on udev for this.
44 * don't rely on the linux symlink in the udev/klibc dir since
45 udev build doesn't use it anymore. This corrects build breakage
46 2004-01-19 multipath-013
47 * update the DM target synthax to the 2.6.0-udm5 style
48 2003-12-29 multipath-012
49 * check hotplug event refers to a block device; if not exit early
51 * add the uninstall target in Makefile
52 2003-12-22 multipath-010
53 * tweak the install target in Makefile
54 * stop passing fds as argument : this change enable a strict
55 segregation of ugly 2.4 code
56 * sysfs version of get_lun_strings()
57 * be careful about the return of get_unique_id() since errors
58 formerly caught up by if(open()) in the caller fn are now returned
60 * send get_serial() in unused.c
61 * introduce dm-simplecmd for RESUME & SUSPEND requests
62 * split add_map() in setup_map() & dm-addmap()
63 * setup_map() correctly submits "SUSPEND-RELOAD-RESUME or CREATE"
64 sequences instead of the bogus "RELOAD or CREATE"
65 * don't print .sg_dev if equal to .dev (2.6) in print_path()
66 * since the kernel code handles defective paths, remove all
67 code to cope with them :
68 * move do_tur() to unused.c
69 * remove .state from path struct
70 * remove .state settings & conditionals
71 * add a cmdline switch to force maps to failover mode,
72 ie 1 path per priority group
73 * add default policies to the whitelist array (spread io ==
74 MULTIBUS / io forced to 1 path == FAILOVER)
75 * move get_disk_size() call out of add_map() to coalesce()
76 * comment tricky coalesce() fn
77 * bogus unsused.c file renamed to unused.c
78 2003-12-20 multipath-010
79 * big ChangeLog update
80 * start to give a little control over target params :
81 introduce cmdline arg -i to control polling interval
82 * cope with hotplug-style calling convention :
83 ie "multipath scsi $DEVPATH" ... to avoid messing with
84 online maps not concerned by an event
85 * example hotplug agent to drop in /etc/hotplug.d/scsi
86 * revert the run & resched patch : unless someone proves me
87 wrong, this was overdesigned
88 * move commented out functions in unused.c
89 * update multipath target params to "udm[23] style"
90 * mp target now supports nr_path == 1, so do we
91 * add gratuitous free()
92 * push version forward
93 2003-12-15 multipath-009
94 * Make the HW-specific get_unique_id switch pretty
95 * Prepare to field-test by whitelisting all known fibre array,
96 try to fetch WWID from the standard EVPD 0x83 off 8 for everyone
97 * configure the multipath target with round-robin path selector and
98 conservative default for a start (udm1 style) :
99 yes it makes this release the firstreally useful one.
100 * temporarily disable map creation for single path device
101 due to current restrictive defaults in the kernel target.
102 Sistina should work it out.
103 * correct the strncmp logic in blacklist function.
104 * update the Makefiles to autodetect libgcc.a & gcc includes
105 "ulibc-style". Factorisation of udevdirs & others niceties
106 * drop a hint about absent /dev/sd? on failed open()
107 * implement a reschedule flag in /var/run.
108 Last thing the prog do before exit is check if a call to multipath
109 was done (but canceled by /var/run/multipath.run check) during its
110 execution. If so restart themain loop.
111 * implement a blacklist of sysfs bdev to not bother with for now
112 (hd,md, dm, sr, scd, ram, raw).
113 This avoid sending SG_IO to unappropiate devices.
114 * Adds a /var/run/multipath.run handling to avoid simultaneous runs.
115 * Remove a commented-out "printf"
116 * drop a libdevmapper copy in extras/multipath;
117 maybe discussions w/Sistina folks will bring a better solution in the future.
118 * drop a putchar usage in libdevmapper to compile cleanly with klibc
119 * drop another such usage of my own in main.c
120 * massage the Makefile to compile libdevmapper against klibc
121 * use "ld" to produce the binary rather than "gcc -static"
122 * stop being stupid w/ uneeded major, minor & dev in main.c:dm_mk_node()
123 * reverse to creating striped target for now because the multipath target
124 is more hairy than expected initialy
125 * push the version code to 009 to be in synch w/ udev
126 2003-11-27 multipath-007
127 * removes sg_err.[ch] deps
128 * makes sure the core code play nice with klibc
129 * port the sysfs calls to dlist helpers
130 * links against udev's sysfs (need libsysfs.a & dlist.a)
131 * finally define DM_TARGET as "multipath" as Joe posted the code today (not tested yet)
132 * push version forward (do you want it in sync with udev version?)
134 * merged in udev-006 tree
135 2003-09-18 Christophe Varoqui <christophe.varoqui@free.fr>
136 * multipath 0.0.1 released.