Closed anchal-agrawal closed 8 years ago
It looks like the 20 second wait before the curl call is still not enough; the output indicates that nginx is still not initialized.
Also seen in https://ci.vmware.run/vmware/vic/5120.
Definitely a timing issue. curl
often times out in this case, and a retry resolves the issue.
Seen again in https://ci.vmware.run/vmware/vic/5224 and https://ci.vmware.run/vmware/vic/5216.
Upon further debugging with @hmahmood, we repro'd this on a local ESX running two VCHs on the default bridge network. Two containers (one of them being nginx) spun up with different VCHs ended up with the same IP and we'd get a connection refused
error upon hitting the nginx endpoint even though we verified that the nginx process was running. Upon removing the second container with the same IP, we can connect to nginx just fine.
The cause for this failure is container IP conflict due to multiple VCHs sharing the same bridge network (which is unsupported). @mhagen-vmware has created #2524 to resolve this in CI.
A recent master build (https://ci.vmware.run/vmware/vic/4863) failed with:
From
log.html
:This is after a
start
on a container created withcreate -it -p 10000:80 -p 10001:80 --name webserver nginx
integration_logs_4863_dbfaf282c1ef3d0cd395bb54ddee6d4e3bb458fd_1474049824.zip