1 2004-02-21 multipath-018
2 * From the Debian SID inclusion review (Philipp Matthias Hahn)
3 * use DESTDIR install prefix in the Makefile
4 * add man pages for devmap_name & multipath
5 * correct libsysfs.h includes
6 * fork the hotplug script in its own shell
7 * Sync with the kernel device mapper code as of 2.6.3-udm3
8 ie. Remove the test interval parameter and its uses
9 * Remove superfluous scsi parameter passed from hotplug
10 * Add the man pages to the [un]install targets
11 2004-02-17 multipath-017
12 * remove the restrictive -f flag.
13 Introduce a more generic "-m iopolicy" one.
14 * remove useless "int with_sysfs" in env struct
15 2004-02-04 multipath-016
16 * add a GROUP_BY_SERIAL flag. This should be useful for
17 controlers that activate they spare paths on simple IO
18 submition with a penalty. The StorageWorks HW defaults to
19 this mode, even if the MULTIBUS mode is OK.
20 * remove unused sg_err.c
21 * big restructuring : split devinfo.c from main.c. Export :
22 * void basename (char *, char *);
23 * int get_serial (int, char *);
24 * int get_lun_strings (char *, char *, char *, char *);
25 * int get_evpd_wwid(char *, char *);
26 * long get_disk_size (char *);
27 * stop passing struct env as param
28 * add devmap_name proggy for udev to name devmaps as per their
29 internal DM name and not only by their sysfs enum name (dm-*)
30 The corresponding udev.rules line is :
31 KERNEL="dm-[0-9]*", PROGRAM="/sbin/devmap_name %M %m", NAME="%k", SYMLINK="%c"
32 * remove make_dm_node fn & call. Rely on udev for this.
33 * don't rely on the linux symlink in the udev/klibc dir since
34 udev build doesn't use it anymore. This corrects build breakage
35 2004-01-19 multipath-013
36 * update the DM target synthax to the 2.6.0-udm5 style
37 2003-12-29 multipath-012
38 * check hotplug event refers to a block device; if not exit early
40 * add the uninstall target in Makefile
41 2003-12-22 multipath-010
42 * tweak the install target in Makefile
43 * stop passing fds as argument : this change enable a strict
44 segregation of ugly 2.4 code
45 * sysfs version of get_lun_strings()
46 * be careful about the return of get_unique_id() since errors
47 formerly caught up by if(open()) in the caller fn are now returned
49 * send get_serial() in unused.c
50 * introduce dm-simplecmd for RESUME & SUSPEND requests
51 * split add_map() in setup_map() & dm-addmap()
52 * setup_map() correctly submits "SUSPEND-RELOAD-RESUME or CREATE"
53 sequences instead of the bogus "RELOAD or CREATE"
54 * don't print .sg_dev if equal to .dev (2.6) in print_path()
55 * since the kernel code handles defective paths, remove all
56 code to cope with them :
57 * move do_tur() to unused.c
58 * remove .state from path struct
59 * remove .state settings & conditionals
60 * add a cmdline switch to force maps to failover mode,
61 ie 1 path per priority group
62 * add default policies to the whitelist array (spread io ==
63 MULTIBUS / io forced to 1 path == FAILOVER)
64 * move get_disk_size() call out of add_map() to coalesce()
65 * comment tricky coalesce() fn
66 * bogus unsused.c file renamed to unused.c
67 2003-12-20 multipath-010
68 * big ChangeLog update
69 * start to give a little control over target params :
70 introduce cmdline arg -i to control polling interval
71 * cope with hotplug-style calling convention :
72 ie "multipath scsi $DEVPATH" ... to avoid messing with
73 online maps not concerned by an event
74 * example hotplug agent to drop in /etc/hotplug.d/scsi
75 * revert the run & resched patch : unless someone proves me
76 wrong, this was overdesigned
77 * move commented out functions in unused.c
78 * update multipath target params to "udm[23] style"
79 * mp target now supports nr_path == 1, so do we
80 * add gratuitous free()
81 * push version forward
82 2003-12-15 multipath-009
83 * Make the HW-specific get_unique_id switch pretty
84 * Prepare to field-test by whitelisting all known fibre array,
85 try to fetch WWID from the standard EVPD 0x83 off 8 for everyone
86 * configure the multipath target with round-robin path selector and
87 conservative default for a start (udm1 style) :
88 yes it makes this release the firstreally useful one.
89 * temporarily disable map creation for single path device
90 due to current restrictive defaults in the kernel target.
91 Sistina should work it out.
92 * correct the strncmp logic in blacklist function.
93 * update the Makefiles to autodetect libgcc.a & gcc includes
94 "ulibc-style". Factorisation of udevdirs & others niceties
95 * drop a hint about absent /dev/sd? on failed open()
96 * implement a reschedule flag in /var/run.
97 Last thing the prog do before exit is check if a call to multipath
98 was done (but canceled by /var/run/multipath.run check) during its
99 execution. If so restart themain loop.
100 * implement a blacklist of sysfs bdev to not bother with for now
101 (hd,md, dm, sr, scd, ram, raw).
102 This avoid sending SG_IO to unappropiate devices.
103 * Adds a /var/run/multipath.run handling to avoid simultaneous runs.
104 * Remove a commented-out "printf"
105 * drop a libdevmapper copy in extras/multipath;
106 maybe discussions w/Sistina folks will bring a better solution in the future.
107 * drop a putchar usage in libdevmapper to compile cleanly with klibc
108 * drop another such usage of my own in main.c
109 * massage the Makefile to compile libdevmapper against klibc
110 * use "ld" to produce the binary rather than "gcc -static"
111 * stop being stupid w/ uneeded major, minor & dev in main.c:dm_mk_node()
112 * reverse to creating striped target for now because the multipath target
113 is more hairy than expected initialy
114 * push the version code to 009 to be in synch w/ udev
115 2003-11-27 multipath-007
116 * removes sg_err.[ch] deps
117 * makes sure the core code play nice with klibc
118 * port the sysfs calls to dlist helpers
119 * links against udev's sysfs (need libsysfs.a & dlist.a)
120 * finally define DM_TARGET as "multipath" as Joe posted the code today (not tested yet)
121 * push version forward (do you want it in sync with udev version?)
123 * merged in udev-006 tree
124 2003-09-18 Christophe Varoqui <christophe.varoqui@free.fr>
125 * multipath 0.0.1 released.