forked from gluster/gstatus
-
Notifications
You must be signed in to change notification settings - Fork 0
/
TODO
102 lines (80 loc) · 4.32 KB
/
TODO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
To Do List / Change Log
0.1 (09 Feb 2014) - Initial Commit
- Show cluster summary state - nodes, volumes and brick up/down state
- show summary information per volume - brick up/down count, volume type,
volume status and usable capacity summary
- show errors discovered in summary list at the end of the output
0.2 Complete 2014/02/24
- align volume state text with nagios monitoring project (DONE)
- add self heal status (DONE)
0.3 Complete 2014/03/01
- add check for self heal backlog and show it in volume summary info (DONE)
- add protocol summary to volume info (DONE)
- reformatted the summary display to make it more readable/les cluttered (DONE)
- fix - in vol status sometimes the node name from gluster is not resolved,
so pick this up and resolve 'manually' (DONE)
0.4 Complete 2014/03/04
- add a version check, to abort the run if gluster is not >= 3.4 (DONE)
- add capacity info to the summary section (DONE)
- add pct used for each volume (DONE)
- add volume detail information (DONE)
- vol components, replica set relationships
- fix - malformed xml from gluster showing bricks against wrong vols
0.45 Complete 2014/03/06
- fix release
- node discovery now validated against initial brick list from a vol info
avoiding name issues (hopefully!)
- gstatus detects whether the cluster is name or ip based, and changes
name resolution logic accordingly
- merged volume methods for calculating self heal states
- added code to check gluster command o/p return code (on vol status) since
--xml does not set shell return code ($?)
- New additions
- added a message count against the cluster state, to immediately show how
many errors a cluster may have i.e. UNHEALTHY(5) --> 5 errors found
0.5 Complete 2014/03/21
- add output options for json, keyvalue and console mode (console is default ie. stdout)
- add client connection counts (glusterfs connections)
0.56
- added debug mode (-D)
- added a config module for global vars across the modules
- updated hostname resolution to use shortname and FQDN
- add code in main gluster module for debug.
0.60 August 2014
- refactor the node definition logic (now based on uuid not name)
- fix - supports cluster defined on fqdn and volumes using shortnames and IP based clusters
- fix - corrected cluster capacity calculations when brick(s) used by multiple volumes
- added capacity overcommit flag if the same device is referenced by multiple bricks
- added overcommit status field added to output (console,keyvalue,json)
- added volume stats to json output mode
- added info message when bricks are missing to confirm the capacity info is inaccurate
- added -n option to skip self heal info gathering
- misc doc updates and refresh of examples directory
0.64
- support added for disperse volumes
0.65 planned
- add task info to the volume state to show when rebalance is active
- provide threshold triggers
- space usage triggers on volumes
- space usage trigger on bricks
0.7 planned
- address the blocking problem when a node goes down (commands to a separate thread with a
timeout?)
0.8 planned
- add brick check to make recommendations on mount point definitions/filesystem, and an option to suppress recommendations
Items being considered
- RFE - Peter AuYeung - for volume capacity, include quota information if enabled
- RFE - Peter AuYeung - for the xml output provide a volume option
- add a highlevel vol type to simplify decision logic
Known BUGS
- invocation immediately after a node down event will block vol status. This needs to be
detected and the run aborted to prevent the wrong information being shown to
the user
- gluster is not returning valid xml when nodes are not present in the config. The problem gets worse with cluster size, and disruption. Current testing has nly allowed 4 node tests to be performed. WORKED AROUND 0.4 onwards
- when multiple replicated volumes are present, the count of self heal processes is incorrect
- for disperse volumes the self heal daemon information is not returned on the vol status command glusterfs 3.7.0
- pending upstream code merge - http://review.gluster.org/10764
update code to use object in all class statement
re-align to 80 char limit
move to spaces not tabs
line 315 - can this error message be more descriptive?