root / doc / design-storagespace.rst @ fb4b885a
History | View | Annotate | Download (4.2 kB)
1 |
============================ |
---|---|
2 |
Storage free space reporting |
3 |
============================ |
4 |
|
5 |
.. contents:: :depth: 4 |
6 |
|
7 |
Background |
8 |
========== |
9 |
|
10 |
Currently Space reporting is broken for all storage types except drbd or |
11 |
lvm (plain). This design looks at the root causes and proposes a way to |
12 |
fix it. |
13 |
|
14 |
Proposed changes |
15 |
================ |
16 |
|
17 |
The changes below will streamline Ganeti to properly support |
18 |
interaction with different storage types. |
19 |
|
20 |
Configuration changes |
21 |
--------------------- |
22 |
|
23 |
Each storage type will have a new "pools" parameter added (type list of |
24 |
strings). This will be list of vgs for plain and drbd (note that we make |
25 |
no distinction at this level between allowed vgs and metavgs), the list |
26 |
of rados pools for rados, or the storage directory for file and |
27 |
sharedfile. The parameters already present in the cluster config object |
28 |
will be moved to the storage parameters. |
29 |
|
30 |
Since currently file and sharedfile only support a single directory this |
31 |
list will be limited to one. In the future if we'll have support for |
32 |
more directories, or for per-nodegroup directories this can be changed. |
33 |
|
34 |
Note that these are just "mechanisms" parameters that define which |
35 |
storage pools the cluster can use. Further filtering about what's |
36 |
allowed can go in the ipolicy, but these changes are not covered in this |
37 |
design doc. |
38 |
|
39 |
Since the ipolicy currently has a list of enabled storage types, we'll |
40 |
use that to decide which storage type is the default, and to self-select |
41 |
it for new instance creations, and reporting. |
42 |
|
43 |
Enabling/disabling of storage types at ``./configure`` time will be |
44 |
eventually removed. |
45 |
|
46 |
RPC changes |
47 |
----------- |
48 |
|
49 |
The noded RPC call that reports node storage space will be changed to |
50 |
accept a list of <method>,<key> string tuples. For each of them it will |
51 |
report the free amount of space found on storage <key> as known by the |
52 |
requested method. Methods are for example ``lvm``, ``filesystem``, |
53 |
``rados``, and the key would be a volume group name in the case of lvm, |
54 |
a directory name for the filesystem and a rados pool name, for |
55 |
rados_pool. |
56 |
|
57 |
Masterd will know (through a constant map) which storage type uses which |
58 |
method for storage calculation (i.e. ``plain`` and ``drbd`` use ``lvm``, |
59 |
``file`` and ``sharedfile`` use ``filesystem``, etc) and query the one |
60 |
needed (or all of the needed ones). |
61 |
|
62 |
Note that for file and sharedfile the node knows which directories are |
63 |
allowed and won't allow any other directory to be queried for security |
64 |
reasons. The actual path still needs to be passed to distinguish the |
65 |
two, as the method will be the same for both. |
66 |
|
67 |
These calculations will be implemented in the node storage system |
68 |
(currently lib/storage.py) but querying will still happen through the |
69 |
``node info`` call, to avoid requiring an extra RPC each time. |
70 |
|
71 |
Ganeti reporting |
72 |
---------------- |
73 |
|
74 |
``gnt-node list`` will by default report information just about the |
75 |
default storage type. It will be possible to add fields to ask about |
76 |
other ones, if they're enabled. |
77 |
|
78 |
``gnt-node info`` will report information about all enabled storage |
79 |
types, without querying them (just say which ones are supported |
80 |
according to the cluster configuration). |
81 |
|
82 |
``gnt-node list-storage`` will change to report information about all |
83 |
available storage pools in each storage type. An extra flag will be |
84 |
added to filter by storage pool name (alternatively we can implement |
85 |
this by allowing to query by a list of ``type:pool`` string tuples to |
86 |
have a more comprehensive filter). |
87 |
|
88 |
|
89 |
Allocator changes |
90 |
----------------- |
91 |
|
92 |
The iallocator protocol doesn't need to change: since we know which |
93 |
storage type an instance has, we'll pass only the "free" value for that |
94 |
storage type to the iallocator, when asking for an allocation to be |
95 |
made. Note that for DRBD nowadays we ignore the case when vg and metavg |
96 |
are different, and we only consider the main VG. Fixing this is outside |
97 |
the scope of this design. |
98 |
|
99 |
Rebalancing changes |
100 |
------------------- |
101 |
|
102 |
Hbal will not need changes, as it handles it already. We don't forecast |
103 |
any changes needed to it. |
104 |
|
105 |
Space reporting changes |
106 |
----------------------- |
107 |
|
108 |
Hspace will by default report by assuming the allocation will happen on |
109 |
the default storage for the cluster/nodegroup. An option will be added |
110 |
to manually specify a different storage. |
111 |
|
112 |
.. vim: set textwidth=72 : |
113 |
.. Local Variables: |
114 |
.. mode: rst |
115 |
.. fill-column: 72 |
116 |
.. End: |