root / NEWS @ 83ad1f3c
History | View | Annotate | Download (7.1 kB)
1 |
Version 0.1.8 (Tue, 29 Sep 2009): |
---|---|
2 |
- Brown-paper-bag release fixing haddock issues |
3 |
|
4 |
Version 0.1.7 (Mon, 28 Sep 2009): |
5 |
- Fixed a bug in the Luxi backend for big responses |
6 |
- Fixed test suite exit code in presence of test failures |
7 |
- Changed the migrate operation to run instead failover for instances |
8 |
which were marked as not running in the input data (this could have |
9 |
been changed since then, but it's better than today's always |
10 |
migrate) |
11 |
- Added support for 'cheap' moves only (only migrate/failover) in |
12 |
balancing |
13 |
- Added support for building without curl (thus no RAPI backend) |
14 |
|
15 |
Version 0.1.6 (Wed, 19 Aug 2009): |
16 |
- Added support for Luxi (the native Ganeti protocol) |
17 |
- Added support for simulated clusters (for hspace only) |
18 |
- Added timeouts for the RAPI backend |
19 |
- Fixed a few inconsistencies in the command line handling |
20 |
- Fixed handling of errors while loading data |
21 |
- The 'network' is a new dependency due to the Luxi addition |
22 |
|
23 |
Version 0.1.5 (Thu, 09 Jul 2009): |
24 |
- Removed obsolete hn1 program; this allowed removal of a lot of |
25 |
supporting code |
26 |
- Lots of changes in hspace: the output now is a shell fragment in |
27 |
order for script to source it or parse it easier; added failure |
28 |
reasons; optimised to use less memory for large clusters |
29 |
- Optimized the scoring algorithm (used by all tools) so that now |
30 |
computations should be faster |
31 |
|
32 |
Version 0.1.4 (Tue, 16 Jun 2009): |
33 |
- Added CPU count/ratio of virtual-to-physical CPUs to the cluster |
34 |
scoring methods; this means that now the balancer, the iallocator |
35 |
plugin and so on will try to keep the VCPU-to-PCPU ratio equal |
36 |
across the cluster |
37 |
- Fixed some hscan bugs |
38 |
- Fixed the way iallocator reads the total disk size (was broken and |
39 |
it was always falling back to summing the disk sizes) |
40 |
- Internals: fixed most compile-time warnings |
41 |
|
42 |
Version 0.1.3 (Fri, 05 Jun 2009): |
43 |
- Fix a bug in the ReplacePrimary instance moves, affecting most of |
44 |
the tools |
45 |
|
46 |
Version 0.1.2 (Tue, 02 Jun 2009): |
47 |
- Add a new program, “hspace”, which computes the free space on a |
48 |
cluster (based on a given instance spec) |
49 |
- Improvements in API docs and partially in the user docs |
50 |
- Started adding unittests |
51 |
|
52 |
Version 0.1.1 (Tue, 26 May 2009): |
53 |
- Add a new program, “hail”, which is an iallocator plugin and can |
54 |
allocate/relocate instances |
55 |
- Experimental support for non-mirrored instances (hail supports them, |
56 |
hbal should no longer abort when it finds such instances and simply |
57 |
ignore them) |
58 |
- The RAPI port and/or scheme can be overriden now, and even “file://” |
59 |
schemes can be used if the message body has been saved under the |
60 |
appropriate name |
61 |
- Lots of code reorganization, esp. rewritten loading pipeline |
62 |
- Better data checking and better error messages in case validation |
63 |
fails; tools now consider nodes with error in input data (‘?’ |
64 |
returned by ganeti) as offline |
65 |
- Small enhancement to the makefile for simpler packaging |
66 |
|
67 |
Version 0.1.0 (Tue, 19 May 2009): |
68 |
- Drop compatibility with Ganeti 1.2 |
69 |
- Add a new minimum score option (with a very low default), should |
70 |
help with very good clusters (but is still not optimal) |
71 |
- Add a --quiet option to hbal |
72 |
- Add support for reading offline nodes directly from the cluster |
73 |
|
74 |
Version 0.0.8 (Tue, 21 Apr 2009): |
75 |
- hbal: prevent mismatches in wrong node names being passed to -O, by |
76 |
aborting in this case |
77 |
- add the ability to write the commands (-C) to a script via |
78 |
(-C<file>), so that it can be later executed directly; this has also |
79 |
changed the commands to include the ncessary -f flags to skip |
80 |
confirmations |
81 |
- add checks for extra argument in hbal and hn1, so that unintended |
82 |
errors are catched |
83 |
- raise the accepted “missing” memory limit to 512MB, to cover usual |
84 |
Xen reservations |
85 |
|
86 |
Version 0.0.7 (Mon, 23 Mar 2009): |
87 |
- added support for offline nodes, which are not used as targets for |
88 |
instance relocation and if they hold instances the hbal algorithm |
89 |
will attempt to relocate these away |
90 |
- added support for offline instances, which now will no longer skew |
91 |
the free memory estimation of nodes; the algorithm will no longer |
92 |
create conditions for N+1 failures when such instances are later |
93 |
started |
94 |
- implemented a complete model of node resources, in order to |
95 |
prevent an unintended re-occurrence of cases like the offline |
96 |
instance were we miscalculate some node resource; this gives warning |
97 |
now in case the node reported free disk or free memory deviates by |
98 |
more than a set amount from the expected value |
99 |
- a new tool *hscan* that can generate the input text-file for the |
100 |
other tools by collection via RAPI |
101 |
- some small changes to the build system to make it more friendly; |
102 |
also included the generated documentation in the source archive |
103 |
|
104 |
Version 0.0.6 (Mon, 16 Mar 2009): |
105 |
- re-factored the hbal algorithm to make it stable in the sense that it |
106 |
gives the same solution when restarted from the middle; barring |
107 |
rounding of disk/memory and incomplete reporting from Ganeti (for |
108 |
1.2), it should be now feasible to rely on its output without |
109 |
generating moves ad infinitum |
110 |
- the hbal algorithm now uses two more variables: the node N+1 |
111 |
failures and the amount of reserved memory; the first of which tries |
112 |
to ‘fix’ the N+1 status, the latter tries to distribute secondaries |
113 |
more equally |
114 |
- the hbal algorithm now uses two more moves at each step: |
115 |
replace+failover and failover+replace (besides the original |
116 |
failover, replace, and failover+replace+failover) |
117 |
- slightly changed the build system to embed GIT version/tags into the |
118 |
binaries so that we know for a binary from which tree it was done, |
119 |
either via ‘--version’ or via “strings hbal|grep version” |
120 |
- changed the solution list and in general the hbal output to be more |
121 |
clear by default, and changed “gnt-instance failover” to |
122 |
“gnt-instance migrate” |
123 |
- added man pages for the two binaries |
124 |
|
125 |
Version 0.0.5 (Mon, 09 Mar 2009): |
126 |
- a few small improvements for hbal (possibly undone by later |
127 |
changes), hbal is now quite faster |
128 |
- fix documentation building |
129 |
- allow hbal to work on non N+1 compliant clusters, but without |
130 |
guarantees that the end cluster will be compliant; in any case, this |
131 |
should give a smaller number of nodes that are not compliant if the |
132 |
cluster state permits it |
133 |
- strip common domain suffix from nodes and instances, so that output |
134 |
is shorter and hopefully clearer |
135 |
|
136 |
Version 0.0.4 (Sun, 15 Feb 2009): |
137 |
- better balancing algorithm in hbal |
138 |
- implemented an RAPI collector, now the cluster data can be gathered |
139 |
automatically via RAPI and doesn't need manual export of node and |
140 |
instance list |
141 |
|
142 |
Version 0.0.3 (Wed, 28 Jan 2009): |
143 |
- initial release of the hbal, a cluster rebalancing tool |
144 |
- input data format changed due to hbal requirements |
145 |
|
146 |
Version 0.0.2 (Tue, 06 Jan 2009): |
147 |
- fix handling of some common cases (cluster N+1 compliant from the |
148 |
start, too big depth given, failure to compute solution) |
149 |
- add option to print the needed command list for reaching the |
150 |
proposed solution |
151 |
|
152 |
Version 0.0.1 (Tue, 06 Jan 2009): |
153 |
- initial release of hn1 tool |