crmsh
crmsh copied to clipboard
ClusterLabs
bootstrap multiple corosync links in the interactive mode enters to the dead loop
Here is the environment.
adm@tw-1:~> sudo ip r
default via 192.168.100.1 dev enp1s0 proto dhcp src 192.168.100.179 metric 101
default via 192.1.160.1 dev enp8s0 proto static metric 20103
default via 192.2.160.1 dev enp9s0 proto static metric 20104
127.0.0.0/8 dev lo proto kernel scope link src 127.0.0.1 metric 30
192.1.160.0/24 dev enp8s0 proto kernel scope link src 192.1.160.101 metric 103
192.2.160.0/24 dev enp9s0 proto kernel scope link src 192.2.160.101 metric 104
192.3.160.0/24 dev enp12s0 proto kernel scope link src 192.3.160.101 metric 100
192.168.0.0/16 dev enp1s0 proto kernel scope link src 192.168.100.179 metric 101
192.168.160.0/24 dev enp7s0 proto kernel scope link src 192.168.160.101 metric 102
adm@tw-1:~>
bootstrap enters a dead loop?
adm@tw-1:~> sudo -E crm cluster init
INFO: Loading "default" profile from /etc/crm/profiles.yml
INFO: Loading "knet-default" profile from /etc/crm/profiles.yml
INFO: Adding public keys to authorized_keys for user root...
INFO: Added public key 2048 SHA256:Ch61e2yqy5Tkw08UEYeNsi9YXBWoujjtYncF+XbuJ3w /home/zzhou/.ssh/id_rsa (RSA).
INFO: A public key is added to authorized_keys for user hacluster: 3072 SHA256:IqRvCzQbjjKiCy5z2IUfJHZO37U+QqhPKNeoh9fIEE0 Cluster internal on tw-1 (RSA)
INFO: Added firewalld service high-availability on tw-1
INFO: Configuring csync2
INFO: Starting csync2.socket service on tw-1
INFO: BEGIN csync2 checking files
INFO: END csync2 checking files
Address for ring0 [192.168.100.179]192.1.160.101
Address for ring1 [192.168.100.179]192.2.160.101
Address for ring2 [192.168.100.179]
Address for ring3 []
Address for ring3 []
Address for ring3 []
Address for ring3 []
Address for ring3 []Ctrl-C, leaving
dead loop, again ...
Address for ring0 [192.168.100.179]192.1.160.101
Address for ring1 [192.168.100.179]
Address for ring2 []
Address for ring2 []
Address for ring2 []
Address for ring2 []
Address for ring2 []
Address for ring3 []Ctrl-C, leaving
However, it turns to the normal situation when configure more links
adm@tw-1:~> sudo -E crm cluster init
INFO: Loading "default" profile from /etc/crm/profiles.yml
INFO: Loading "knet-default" profile from /etc/crm/profiles.yml
INFO: Adding public keys to authorized_keys for user root...
INFO: Added public key 2048 SHA256:Ch61e2yqy5Tkw08UEYeNsi9YXBWoujjtYncF+XbuJ3w /home/zzhou/.ssh/id_rsa (RSA).
INFO: A public key is added to authorized_keys for user hacluster: 3072 SHA256:IqRvCzQbjjKiCy5z2IUfJHZO37U+QqhPKNeoh9fIEE0 Cluster internal on tw-1 (RSA)
INFO: Added firewalld service high-availability on tw-1
INFO: Configuring csync2
INFO: Starting csync2.socket service on tw-1
INFO: BEGIN csync2 checking files
INFO: END csync2 checking files
Address for ring0 [192.168.100.179]192.1.160.101
Address for ring1 [192.168.100.179]192.2.160.101
Address for ring2 [192.168.100.179]192.3.160.101
Address for ring3 [192.168.100.179]192.168.160.101
Address for ring4 [192.168.100.179]
INFO: Configure SBD:
hoops, not really, dead loop, again
Address for ring0 [192.168.100.179]
Address for ring1 []192.1.160.101
Address for ring2 []192.2.160.101
Address for ring3 []192.3.160.101
Address for ring4 []
Address for ring4 []
Address for ring4 []
Address for ring3 []Ctrl-C, leaving
One of reason is the existing crm.conf jeopardize the bootstrap init procedure.
adm@tw-1:~> sudo grep force /root/.config/crm/crm.conf
force = true
BTW, even with force = true in crm.conf, one of the possible user experience improvement could be to print those auto-confirmation during the process as shown below. That could be the possible indicator/hint for force= option in crm.conf.
csync2 is already configured - overwrite (y/n)? y
/etc/corosync/authkey already exists - overwrite (y/n)? y
/etc/corosync/corosync.conf already exists - overwrite (y/n)? y
Add another ring (y/n)? y
Should we stop saving force in crm.conf? I think it should apply only to currrent running command, and should never be saved.
