-
Notifications
You must be signed in to change notification settings - Fork 3
/
Copy pathShutdown and Startup Notes
54 lines (40 loc) · 1.62 KB
/
Shutdown and Startup Notes
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
#!!!WARNING!!!
# These are just my notes, and or spesific to my setup...
# This is also not well tested.
# See official docs:
# https://access.redhat.com/solutions/2139301
#If you are on proxmox, or not using "hyperconverged" nodes, you can probably skip this:
#Stop all non-ceph dockers
# - Exclude LBs
# - Exclude DNS
#Stage 2:
#Get Ceph clients:
ceph tell mds.0 client ls > /dev/shm/tmp.CephClients 2> /dev/null
#Look at this output and make sure you can tolerate kicking the clients!
cat /dev/shm/tmp.CephClients
#Try to umount all ceph clients.
#Clients=`cat /dev/shm/tmp.CephClients | grep '"hostname":' | cut -d '"' -f 4`
#Might re-run the 'ceph tell mds.0 client ls' command to see how it looks, but we will need to kick and stuck clients:
#Evict all other clients:
#ClientIDs=`cat /dev/shm/tmp.CephClients | grep '"id":' | tr -s ' ' | cut -d ' ' -f 3 | cut -d ',' -f 1`
#For each:
#ceph tell mds.0 client evict id=$ClientID
#ceph tell mds.0 client evict client_metadata.=$ClientID
#ceph osd set noout
#ceph osd set norecover
#ceph osd set norebalance
#ceph osd set nobackfill
#ceph osd set nodown
#ceph osd set pause
#You SHOULD be safe to shutdown now....
#Restart time.
#WAIT!!! Really, like watch ceph -s, and wait till it stops changing, THEN wait 5 more minutes. 15 If you have the time! No joke, this is an issue!
#ceph osd unset norecover
#WAIT!!! Watch 'watch ceph -s', and wait till it stops changing
#ceph osd unset nobackfill
#WAIT!!! Watch 'watch ceph -s', and wait till it stops changing
#Now we can really start things up.
#ceph osd unset noout
#ceph osd unset norebalance
#ceph osd unset nodown
#ceph osd unset pause