9d1911c5a4
Bug #1549753 raises false alarm about haproxy to the operator and monitoring tools even though haproxy is running fine and there is no issue. haproxy uses UDP datagram socket for logging and the default queue length of 10 datagrams causes false alert messages in haproxy logs. This patch set fixes this bug by increasing the queue length to 128. Test results of the fix in this patch set are at http://paste.openstack.org/show/490302/. Change-Id: I64fdd17cd278d51616aa0dfa4e8d2a422c8ee630 Closes-Bug: #1549753
18 lines
519 B
Bash
18 lines
519 B
Bash
#!/bin/bash
|
|
|
|
# Set datagram receiving queue length to 128 for haproxy's socket
|
|
sysctl net.unix.max_dgram_qlen=128
|
|
|
|
# We are intentionally not using exec so we can reload the haproxy config later
|
|
echo "Running command: '${CMD}'"
|
|
$CMD
|
|
|
|
# TODO(SamYaple): This has the potential for a race condition triggered by a
|
|
# config reload that could cause the container to exit
|
|
while [[ -e "/proc/$(cat /run/haproxy.pid)" ]]; do
|
|
sleep 5
|
|
done
|
|
|
|
# Based on the above loop this point should never be reached
|
|
exit 1
|