Monitor - reweight near full osd autonomicly » History » Version 1
Jessica Mack, 07/03/2015 08:33 PM
1 | 1 | Jessica Mack | h1. Monitor - reweight near full osd autonomicly |
---|---|---|---|
2 | |||
3 | h3. Summary |
||
4 | |||
5 | Monitor can check osd whose device usage ratio is near full(default ratio 85%), and report "health |
||
6 | err". There is usually one device near full while the others not, so monitor can simply decrease |
||
7 | weight of the near full osd to migrate data to the other osds, which can delay the ratio increase. |
||
8 | This is usefull for administor to extend cluster capcity, especially in a large scale osd cluster. |
||
9 | |||
10 | h3. Owners |
||
11 | |||
12 | * Min Chen (minchen@ubuntukylin.com) |
||
13 | * Name (Affiliation) |
||
14 | * Name |
||
15 | |||
16 | h3. Interested Parties |
||
17 | |||
18 | * Name (Affiliation) |
||
19 | * Name (Affiliation) |
||
20 | * Name |
||
21 | |||
22 | h3. Current Status |
||
23 | |||
24 | Administor can reweight a near full osd by command. If a series of near full osds |
||
25 | occur in a long time, then the maintain work is heavy. |
||
26 | Detailed Description |
||
27 | The monitor get the log "near full" from the heavy load osd by LogMonitor, |
||
28 | and we can let it to reweight osd. |
||
29 | |||
30 | h3. Work items |
||
31 | |||
32 | h4. Coding tasks |
||
33 | |||
34 | # Task 1 |
||
35 | # Task 2 |
||
36 | # Task 3 |
||
37 | |||
38 | h4. Build / release tasks |
||
39 | |||
40 | # Task 1 |
||
41 | # Task 2 |
||
42 | # Task 3 |
||
43 | |||
44 | h4. Documentation tasks |
||
45 | |||
46 | # Task 1 |
||
47 | # Task 2 |
||
48 | # Task 3 |
||
49 | |||
50 | h4. Deprecation tasks |
||
51 | |||
52 | # Task 1 |
||
53 | # Task 2 |
||
54 | # Task 3 |