root / NEWS @ 050b0c38
History | View | Annotate | Download (3.3 kB)
1 |
Version 0.0.7 (Mon, 23 Mar 2009): |
---|---|
2 |
- added support for offline nodes, which are not used as targets for |
3 |
instance relocation and if they hold instances the hbal algorithm |
4 |
will attempt to relocate these away |
5 |
- added support for offline instances, which now will no longer skew |
6 |
the free memory estimation of nodes; the algorithm will no longer |
7 |
create conditions for N+1 failures when such instances are later |
8 |
started |
9 |
- implemented a complete model of node resources, in order to |
10 |
prevent an unintended re-occurrence of cases like the offline |
11 |
instance were we miscalculate some node resource; this gives warning |
12 |
now in case the node reported free disk or free memory deviates by |
13 |
more than a set amount from the expected value |
14 |
- a new tool *hscan* that can generate the input text-file for the |
15 |
other tools by collection via RAPI |
16 |
- some small changes to the build system to make it more friendly; |
17 |
also included the generated documentation in the source archive |
18 |
|
19 |
Version 0.0.6 (Mon, 16 Mar 2009): |
20 |
- re-factored the hbal algorithm to make it stable in the sense that it |
21 |
gives the same solution when restarted from the middle; barring |
22 |
rounding of disk/memory and incomplete reporting from Ganeti (for |
23 |
1.2), it should be now feasible to rely on its output without |
24 |
generating moves ad infinitum |
25 |
- the hbal algorithm now uses two more variables: the node N+1 |
26 |
failures and the amount of reserved memory; the first of which tries |
27 |
to ‘fix’ the N+1 status, the latter tries to distribute secondaries |
28 |
more equally |
29 |
- the hbal algorithm now uses two more moves at each step: |
30 |
replace+failover and failover+replace (besides the original |
31 |
failover, replace, and failover+replace+failover) |
32 |
- slightly changed the build system to embed GIT version/tags into the |
33 |
binaries so that we know for a binary from which tree it was done, |
34 |
either via ‘--version’ or via “strings hbal|grep version” |
35 |
- changed the solution list and in general the hbal output to be more |
36 |
clear by default, and changed “gnt-instance failover” to |
37 |
“gnt-instance migrate” |
38 |
- added man pages for the two binaries |
39 |
|
40 |
Version 0.0.5 (Mon, 09 Mar 2009): |
41 |
- a few small improvements for hbal (possibly undone by later |
42 |
changes), hbal is now quite faster |
43 |
- fix documentation building |
44 |
- allow hbal to work on non N+1 compliant clusters, but without |
45 |
guarantees that the end cluster will be compliant; in any case, this |
46 |
should give a smaller number of nodes that are not compliant if the |
47 |
cluster state permits it |
48 |
- strip common domain suffix from nodes and instances, so that output |
49 |
is shorter and hopefully clearer |
50 |
|
51 |
Version 0.0.4 (Sun, 15 Feb 2009): |
52 |
- better balancing algorithm in hbal |
53 |
- implemented an RAPI collector, now the cluster data can be gathered |
54 |
automatically via RAPI and doesn't need manual export of node and |
55 |
instance list |
56 |
|
57 |
Version 0.0.3 (Wed, 28 Jan 2009): |
58 |
- initial release of the hbal, a cluster rebalancing tool |
59 |
- input data format changed due to hbal requirements |
60 |
|
61 |
Version 0.0.2 (Tue, 06 Jan 2009): |
62 |
- fix handling of some common cases (cluster N+1 compliant from the |
63 |
start, too big depth given, failure to compute solution) |
64 |
- add option to print the needed command list for reaching the |
65 |
proposed solution |
66 |
|
67 |
Version 0.0.1 (Tue, 06 Jan 2009): |
68 |
- initial release of hn1 tool |