root / NEWS @ a097adcc
History | View | Annotate | Download (4.9 kB)
1 |
Version 0.1.1 (Tue, 26 May 2009): |
---|---|
2 |
- Add a new program, “hail”, which is an iallocator plugin and can |
3 |
allocate/relocate instances |
4 |
- Experimental support for non-mirrored instances (hail supports them, |
5 |
hbal should no longer abort when it finds such instances and simply |
6 |
ignore them) |
7 |
- The RAPI port and/or scheme can be overriden now, and even “file://” |
8 |
schemes can be used if the message body has been saved under the |
9 |
appropriate name |
10 |
- Lots of code reorganization, esp. rewritten loading pipeline |
11 |
- Better data checking and better error messages in case validation |
12 |
fails; tools now consider nodes with error in input data (‘?’ |
13 |
returned by ganeti) as offline |
14 |
- Small enhancement to the makefile for simpler packaging |
15 |
|
16 |
Version 0.1.0 (Tue, 19 May 2009): |
17 |
- Drop compatibility with Ganeti 1.2 |
18 |
- Add a new minimum score option (with a very low default), should |
19 |
help with very good clusters (but is still not optimal) |
20 |
- Add a --quiet option to hbal |
21 |
- Add support for reading offline nodes directly from the cluster |
22 |
|
23 |
Version 0.0.8 (Tue, 21 Apr 2009): |
24 |
- hbal: prevent mismatches in wrong node names being passed to -O, by |
25 |
aborting in this case |
26 |
- add the ability to write the commands (-C) to a script via |
27 |
(-C<file>), so that it can be later executed directly; this has also |
28 |
changed the commands to include the ncessary -f flags to skip |
29 |
confirmations |
30 |
- add checks for extra argument in hbal and hn1, so that unintended |
31 |
errors are catched |
32 |
- raise the accepted “missing” memory limit to 512MB, to cover usual |
33 |
Xen reservations |
34 |
|
35 |
Version 0.0.7 (Mon, 23 Mar 2009): |
36 |
- added support for offline nodes, which are not used as targets for |
37 |
instance relocation and if they hold instances the hbal algorithm |
38 |
will attempt to relocate these away |
39 |
- added support for offline instances, which now will no longer skew |
40 |
the free memory estimation of nodes; the algorithm will no longer |
41 |
create conditions for N+1 failures when such instances are later |
42 |
started |
43 |
- implemented a complete model of node resources, in order to |
44 |
prevent an unintended re-occurrence of cases like the offline |
45 |
instance were we miscalculate some node resource; this gives warning |
46 |
now in case the node reported free disk or free memory deviates by |
47 |
more than a set amount from the expected value |
48 |
- a new tool *hscan* that can generate the input text-file for the |
49 |
other tools by collection via RAPI |
50 |
- some small changes to the build system to make it more friendly; |
51 |
also included the generated documentation in the source archive |
52 |
|
53 |
Version 0.0.6 (Mon, 16 Mar 2009): |
54 |
- re-factored the hbal algorithm to make it stable in the sense that it |
55 |
gives the same solution when restarted from the middle; barring |
56 |
rounding of disk/memory and incomplete reporting from Ganeti (for |
57 |
1.2), it should be now feasible to rely on its output without |
58 |
generating moves ad infinitum |
59 |
- the hbal algorithm now uses two more variables: the node N+1 |
60 |
failures and the amount of reserved memory; the first of which tries |
61 |
to ‘fix’ the N+1 status, the latter tries to distribute secondaries |
62 |
more equally |
63 |
- the hbal algorithm now uses two more moves at each step: |
64 |
replace+failover and failover+replace (besides the original |
65 |
failover, replace, and failover+replace+failover) |
66 |
- slightly changed the build system to embed GIT version/tags into the |
67 |
binaries so that we know for a binary from which tree it was done, |
68 |
either via ‘--version’ or via “strings hbal|grep version” |
69 |
- changed the solution list and in general the hbal output to be more |
70 |
clear by default, and changed “gnt-instance failover” to |
71 |
“gnt-instance migrate” |
72 |
- added man pages for the two binaries |
73 |
|
74 |
Version 0.0.5 (Mon, 09 Mar 2009): |
75 |
- a few small improvements for hbal (possibly undone by later |
76 |
changes), hbal is now quite faster |
77 |
- fix documentation building |
78 |
- allow hbal to work on non N+1 compliant clusters, but without |
79 |
guarantees that the end cluster will be compliant; in any case, this |
80 |
should give a smaller number of nodes that are not compliant if the |
81 |
cluster state permits it |
82 |
- strip common domain suffix from nodes and instances, so that output |
83 |
is shorter and hopefully clearer |
84 |
|
85 |
Version 0.0.4 (Sun, 15 Feb 2009): |
86 |
- better balancing algorithm in hbal |
87 |
- implemented an RAPI collector, now the cluster data can be gathered |
88 |
automatically via RAPI and doesn't need manual export of node and |
89 |
instance list |
90 |
|
91 |
Version 0.0.3 (Wed, 28 Jan 2009): |
92 |
- initial release of the hbal, a cluster rebalancing tool |
93 |
- input data format changed due to hbal requirements |
94 |
|
95 |
Version 0.0.2 (Tue, 06 Jan 2009): |
96 |
- fix handling of some common cases (cluster N+1 compliant from the |
97 |
start, too big depth given, failure to compute solution) |
98 |
- add option to print the needed command list for reaching the |
99 |
proposed solution |
100 |
|
101 |
Version 0.0.1 (Tue, 06 Jan 2009): |
102 |
- initial release of hn1 tool |