parallelvirtualcluster
/
pvc
1
0
Fork 0
Code Issues 15 Pull Requests Projects Releases 106 Wiki Activity

Modify printout during keepalive to properly handle new daemon/domain state split

This commit is contained in:
Joshua Boniface 2018-06-11 19:04:59 -04:00
parent 7279b66dc2
commit 0f74c68ae0
1 changed files with 19 additions and 17 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
NodeInstance.py
View File

@ -260,24 +260,26 @@ class NodeInstance():
# Update our local node lists # Update our local node lists
for node_name in self.t_node: for node_name in self.t_node:
try: try:
node_state = self.zk.get('/nodes/{}/daemonstate'.format(node_name))[0].decode('ascii') node_daemon_state = self.zk.get('/nodes/{}/daemonstate'.format(node_name))[0].decode('ascii')
node_domain_state = self.zk.get('/nodes/{}/domainstate'.format(node_name))[0].decode('ascii')
node_keepalive = int(self.zk.get('/nodes/{}/keepalive'.format(node_name))[0].decode('ascii')) node_keepalive = int(self.zk.get('/nodes/{}/keepalive'.format(node_name))[0].decode('ascii'))
except: except:
node_state = 'unknown' node_daemon_state = 'unknown'
node_domain_state = 'unknown'
node_keepalive = 0 node_keepalive = 0
# Handle deadtime and fencng if needed # Handle deadtime and fencng if needed
# (A node is considered dead when its keepalive timer is >6*keepalive_interval seconds # (A node is considered dead when its keepalive timer is >6*keepalive_interval seconds
# out-of-date while in 'start' state) # out-of-date while in 'start' state)
node_deadtime = int(time.time()) - ( int(self.config['keepalive_interval']) * 6 ) node_deadtime = int(time.time()) - ( int(self.config['keepalive_interval']) * 6 )
if node_keepalive < node_deadtime and node_state == 'start': if node_keepalive < node_deadtime and node_daemon_state == 'start':
ansiiprint.echo('Node {} is dead - performing fence operation in 3 seconds'.format(node_name), '', 'w') ansiiprint.echo('Node {} seems dead - starting monitor for fencing'.format(node_name), '', 'w')
self.zk.set('/nodes/{}/daemonstate'.format(node_name), 'dead'.encode('ascii')) self.zk.set('/nodes/{}/daemonstate'.format(node_name), 'dead'.encode('ascii'))
fence_thread = threading.Thread(target=fencenode.fence, args=(node_name, self.zk), kwargs={}) fence_thread = threading.Thread(target=fencenode.fence, args=(node_name, self.zk), kwargs={})
fence_thread.start() fence_thread.start()
# Update the arrays # Update the arrays
if node_state == 'start' and node_name not in self.active_node_list: if node_daemon_state == 'start' and node_domain_state != 'flush' and node_name not in self.active_node_list:
self.active_node_list.append(node_name) self.active_node_list.append(node_name)
try: try:
self.flushed_node_list.remove(node_name) self.flushed_node_list.remove(node_name)
@ -287,17 +289,7 @@ class NodeInstance():
self.inactive_node_list.remove(node_name) self.inactive_node_list.remove(node_name)
except ValueError: except ValueError:
pass pass
if node_state == 'flush' and node_name not in self.flushed_node_list: if node_daemon_state != 'start' and node_daemon_state != 'flush' and node_name not in self.inactive_node_list:
self.flushed_node_list.append(node_name)
try:
self.active_node_list.remove(node_name)
except ValueError:
pass
try:
self.inactive_node_list.remove(node_name)
except ValueError:
pass
if node_state != 'start' and node_state != 'flush' and node_name not in self.inactive_node_list:
self.inactive_node_list.append(node_name) self.inactive_node_list.append(node_name)
try: try:
self.active_node_list.remove(node_name) self.active_node_list.remove(node_name)
@ -307,9 +299,19 @@ class NodeInstance():
self.flushed_node_list.remove(node_name) self.flushed_node_list.remove(node_name)
except ValueError: except ValueError:
pass pass
if node_domain_state == 'flush' and node_name not in self.flushed_node_list:
self.flushed_node_list.append(node_name)
try:
self.active_node_list.remove(node_name)
except ValueError:
pass
try:
self.inactive_node_list.remove(node_name)
except ValueError:
pass
# Display cluster information to the terminal # Display cluster information to the terminal
ansiiprint.echo('{}Cluster status{}'.format(ansiiprint.purple(), ansiiprint.end()), '', 't') ansiiprint.echo('{}Cluster status{}'.format(ansiiprint.purple(), ansiiprint.end()), '', 't')
ansiiprint.echo('{}Active nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.active_node_list)), '', 'c') ansiiprint.echo('{}Active nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.active_node_list)), '', 'c')
ansiiprint.echo('{}Flushed nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.flushed_node_list)), '', 'c')
ansiiprint.echo('{}Inactive nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.inactive_node_list)), '', 'c') ansiiprint.echo('{}Inactive nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.inactive_node_list)), '', 'c')
ansiiprint.echo('{}Flushed nodes:{} {}'.format(ansiiprint.bold(), ansiiprint.end(), ' '.join(self.flushed_node_list)), '', 'c')