[Linux-HA] connection to quorumd timing problem?

[Sebastian Reitenbach]

Hi list,

I am still fiddling around with the quorumd. I have quorumd configured in ha.cf
and have a 
quorumd.conf and the environment variable set. I think there must be a timing
problem 
because I started the hearbeat on one of my nodes, where the quorumd runs, so
connection 
to quorumd is more or less, localhost.

At the first start the node came up, without finding its quorum server.  killing
all 
processes and restarting, it had no problem to contact the quorum server.

I saw this line in the logs, both times starting up the quorumd process:

Feb 23 20:57:52 ppsbackup101 heartbeat: [10828]: info: Starting child client
"/usr/lib64/heartbeat/quorumd" (0,0)

on the first try, some seconds later I read the following in the logs, and the
cluster 
node started up without quorum:

Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: quorum plugin: quorumd,
quorumd_init()
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: quorum plugin:
cluster:ppscluster, quorum_server:quorum
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: quorum plugin: quorumd
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: cluster:ppscluster,
member_count=1, member_quorum_votes=100
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: total_node_count=5,
total_quorum_votes=500
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: quorum plugin: quorumd,
connect_quorum_server
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: debug: zhenh: return cur_quorum  -1
Feb 23 20:57:52 ppsbackup101 stonithd: [10837]: info: Signing in with heartbeat.
Feb 23 20:57:52 ppsbackup101 ccm: [10834]: info: G_main_add_SignalHandler: Added
signal handler for signal 15
Feb 23 20:57:52 ppsbackup101 quorumd: [10841]: info: G_main_add_SignalHandler:
Added signal handler for signal 10
Feb 23 20:57:52 ppsbackup101 quorumd: [10841]: info: G_main_add_SignalHandler:
Added signal handler for signal 12
Feb 23 20:57:52 ppsbackup101 quorumd: [10841]: info: G_main_add_SignalHandler:
Added signal handler for signal 1

after killing all processes, and restarting, I saw this in the logs, and the
node came up 
with quorum:


Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: quorum plugin: quorumd,
quorumd_init()
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: quorum plugin:
cluster:ppscluster, quorum_server:quorum
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: quorum plugin: quorumd
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: cluster:ppscluster,
member_count=1, member_quorum_votes=100
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: total_node_count=5,
total_quorum_votes=500
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: quorum plugin: quorumd,
connect_quorum_server
Feb 23 21:02:47 ppsbackup101 quorumd: [10928]: debug: quorumd plugin 2.0.8,
init()
Feb 23 21:02:47 ppsbackup101 quorumd: [10928]: info: load config file
/etc/ha.d/quorumd.conf
Feb 23 21:02:47 ppsbackup101 quorumd: [10928]: debug: create new client 1
Feb 23 21:02:47 ppsbackup101 ccm: [10921]: debug: zhenh: return cur_quorum  1


Do I can do sth. to make the startup more predictable?

kind regards
Sebastian