Since the reload is now synchronous over the master CLI, try to reload
with it. This was a problem before with the signals because it wasn't
possible to wait for the end of the reload before sending the requests.
This activate again this test, we will see if it's more stable or we
will deactivate it again..
varnishtest "Seamless reload issue with abns sockets"
feature ignore_unknown_macro
+feature cmd "command -v socat"
+feature cmd "command -v grep"
# abns@ sockets are not available on freebsd
#EXCLUDE_TARGETS=freebsd,osx,generic
-#REGTEST_TYPE=broken
+#REGTEST_TYPE=devel
-haproxy h1 -W -conf {
+haproxy h1 -W -S -conf {
global
stats socket "${tmpdir}/h1/stats" level admin expose-fd listeners
} -start
shell {
- kill -USR2 $(cat "${tmpdir}/h1/pid")
+ echo "reload" | socat -t1000 TCP:${h1_mcli_addr}:${h1_mcli_port} - | grep 'Success=1'
}
client c1 -connect ${h1_testme_sock} {