parallelvirtualcluster/pvc
1
0
Fork 0
Code Issues 11 Pull Requests Projects Releases 115 Wiki Activity

Adjust fence failcount limit to 6 (30s)

Browse Source 
The previous saving throw limit (3/15s) seems to have been too low. I
was observing bizarre failures where a node would be fenced while it was
still starting up. Some of this may have been related to Zookeeper
connections taking too long, but this was inconsistent.

Increase this to 6 saving throws (30s). This provides significantly more
time for a node to properly check in on startup before another node
fences it. In the real world, 15s vs 30s isn't that big of a downtime
change, but prevents false-positive fences.
This commit is contained in:
Joshua Boniface
2020-08-05 22:36:28 -04:00
parent 02343079c0
commit ccee124c8b
1 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
node-daemon/pvcnoded/fencing.py
View File

@@ -31,9 +31,10 @@ import pvcnoded.VMInstance as VMInstance
# Fence thread entry function # Fence thread entry function
# #
def fenceNode(node_name, zk_conn, config, logger): def fenceNode(node_name, zk_conn, config, logger):
# We allow exactly 6 saving throws (30 seconds) for the host to come back online or we kill it
failcount_limit = 6
failcount = 0 failcount = 0
# We allow exactly 3 saving throws for the host to come back online while failcount < failcount_limit:
while failcount < 3:
# Wait 5 seconds # Wait 5 seconds
time.sleep(5) time.sleep(5)
# Get the state # Get the state
@@ -41,7 +42,7 @@ def fenceNode(node_name, zk_conn, config, logger):
# Is it still 'dead' # Is it still 'dead'
if node_daemon_state == 'dead': if node_daemon_state == 'dead':
failcount += 1 failcount += 1
logger.out('Node "{}" failed {} saving throws'.format(node_name, failcount), state='w') logger.out('Node "{}" failed {}/{} saving throws'.format(node_name, failcount, failcount_limit), state='w')
# It changed back to something else so it must be alive # It changed back to something else so it must be alive
else: else:
logger.out('Node "{}" passed a saving throw; canceling fence'.format(node_name), state='o') logger.out('Node "{}" passed a saving throw; canceling fence'.format(node_name), state='o')