RFE: Consider DLM (dlm_tool) in crmsh

[jirib]

Not sure how helpful it would be but if we would consider crmsh as main cluster tool, then it might be useful.

An example:

$ dlm_tool ls
dlm lockspaces
name          lvm_clvg0
id            0x45d1d4f1
flags         0x00000000
change        member 1 joined 1 remove 0 failed 0 seq 1,1
members       1

name          lvm_global
id            0x12aabd2d
flags         0x00000000
change        member 2 joined 1 remove 0 failed 0 seq 2,2
members       1 2

After adding "shared" VG, that is one which is activated on both nodes (eg. for OCFS2), this happens:

$ crm configure show clvg0
primitive clvg0 LVM-activate \
        params vgname=clvg0 vg_access_mode=lvmlockd activation_mode=shared \
        op start timeout=90s interval=0 \
        op stop timeout=90s interval=0 \
        op monitor interval=90s timeout=90s

# see lvm_clvg0 has two member now!

$ dlm_tool ls
dlm lockspaces
name          lvm_clvg0
id            0x45d1d4f1
flags         0x00000000
change        member 2 joined 1 remove 0 failed 0 seq 1,1
members       1 2

name          lvm_global
id            0x12aabd2d
flags         0x00000000
change        member 2 joined 1 remove 0 failed 0 seq 1,1
members       1 2

# see that we have 'rrp_mode = "passive"', and 'protocol'

$ dlm_tool dump
5674 config file log_debug = 1 cli_set 0 use 1
5674 dlm_controld 4.1.0 started
5674 our_nodeid 1
5674 node_config 1
5674 node_config 2
5674 found /dev/misc/dlm-control minor 124
5674 found /dev/misc/dlm-monitor minor 123
5674 found /dev/misc/dlm_plock minor 122
5674 /sys/kernel/config/dlm/cluster/comms: opendir failed: 2
5674 /sys/kernel/config/dlm/cluster/spaces: opendir failed: 2
5674 set log_debug 1
5674 set mark 0
5674 cmap totem.rrp_mode = 'passive'
5674 set protocol 1
5674 set /proc/sys/net/core/rmem_default 4194304
5674 set /proc/sys/net/core/rmem_max 4194304
5674 set recover_callbacks 1
5674 cmap totem.cluster_name = 'jb155sapqe'
5674 set cluster_name jb155sapqe
5674 /dev/misc/dlm-monitor fd 13
5674 cluster quorum 1 seq 270 nodes 2
5674 cluster node 1 added seq 270
5674 set_configfs_node 1 192.168.253.100 local 1 mark 0
5674 cluster node 2 added seq 270
5674 set_configfs_node 2 192.168.253.101 local 0 mark 0
5674 cpg_join dlm:controld ...
5674 setup_cpg_daemon 15
5674 dlm:controld conf 1 1 0 memb 1 join 1 left 0
5674 daemon joined 1
5674 dlm:controld ring 1:270 2 memb 1 2
5674 receive_protocol 1 max 3.1.1.0 run 0.0.0.0
5674 daemon node 1 prot max 0.0.0.0 run 0.0.0.0
5674 daemon node 1 save max 3.1.1.0 run 0.0.0.0
5674 set_protocol member_count 1 propose daemon 3.1.1 kernel 1.1.1
5674 receive_protocol 1 max 3.1.1.0 run 3.1.1.0
5674 daemon node 1 prot max 3.1.1.0 run 0.0.0.0
5674 daemon node 1 save max 3.1.1.0 run 3.1.1.0
5674 run protocol from nodeid 1
5674 daemon run 3.1.1 max 3.1.1 kernel run 1.1.1 max 1.1.1
5674 plocks 16
5674 receive_protocol 1 max 3.1.1.0 run 3.1.1.0
5674 send_fence_clear 1 fipu
5674 receive_fence_clear from 1 for 1 result -61 flags 1
5674 fence_in_progress_unknown 0 all_fipu
5815 dlm:controld conf 2 1 0 memb 1 2 join 2 left 0
5815 daemon joined 2
5815 receive_protocol 2 max 3.1.1.0 run 0.0.0.0
5815 daemon node 2 prot max 0.0.0.0 run 0.0.0.0
5815 daemon node 2 save max 3.1.1.0 run 0.0.0.0
5815 receive_protocol 1 max 3.1.1.0 run 3.1.1.0
5815 receive_fence_clear from 1 for 2 result 0 flags 6
5815 receive_protocol 2 max 3.1.1.0 run 3.1.1.0
5815 daemon node 2 prot max 3.1.1.0 run 0.0.0.0
5815 daemon node 2 save max 3.1.1.0 run 3.1.1.0

[Roger Zhou]

Hmm, I see what you mean. However, DLM isn't versatile enough for the pacemaker/corosync cluster stack. Out of my head, I lack the suitable user interface to integrate it.