User another TLD instead of .local

[LiAh Sheep]

Let’s change *.openshift.local to *.openshift.vm or whatever:

• cnpaas-openshift/Vagrantfile
• cnpaas-openshift/bin/init/dev-bootstrap.sh
• cnpaas-openshift/bin/init/ext-domain-setup
• cnpaas-openshift/bin/init/set-env
• cnpaas-portal/config/autoload/app_config.env.yml
• cnpaas-portal/config/autoload/paas_cfg.env.yml

---

Some references articles explaining this stupid behaviour:

ACCESSING .LOCAL DOMAINS IS SLOW ON MAC OS X

This appears to be a fault in Mac OS X Lion, where it takes 5 seconds to resolve .local domain names, even if they are hardcoded in your /etc/hosts file. You can work around this problem by using Charles's DNS Spoofing tool intead of /etc/hosts, however you'll experience the .local DNS resolution delay when not using Charles. I recommend using an alternative TLD such as .dev.

• http://www.charlesproxy.com/documentation/faqs/accessing-local-domains-is-slow-on-mac-os/
• http://stackoverflow.com/questions/6841421/mac-osx-lion-dns-lookup-order
• http://www.macwindows.com/Centrify-fix-slow-lion-access-to-SMB-home-directories.html

[LiAh Sheep]

You may have to update the dnsmasq settings to openshift.vm after this merge.

[LiAh Sheep]

The hostname change didn't work because the following line was unstable for Vagrantfile:

config.vm.hostname = "broker.openshift.vm"

Looking for other methods.

[LiAh Sheep]

fix: change .openshift.local to .openshift.vm (fixing issue #99)

→ <<cset 6ad77d41b81b>>

[LiAh Sheep]

Revert "fix: change .openshift.local to .openshift.vm (fixing issue #99)"

This reverts commit 6ad77d41b81b8bc44073d675635a6a1de79fa704.

→ <<cset 9afb24f6f5bb>>

[LiAh Sheep]

The line was unstable because I haven’t updated /etc/hosts properly. This solution should actually be fine.

[Jimmy Chu]

fix: change .openshift.local to .openshift.vm (fixing issue #99)

→ <<cset 3544acbcabcf>>

[Jimmy Chu]

Merged in liahsheep/cnpaas-portal/openshift_vm_hostname (pull request #120)

fix: change .openshift.local to .openshift.vm (fixing issue #99)

→ <<cset 9d630b6dfb61>>

[LiAh Sheep]

Revert "fix: change .openshift.local to .openshift.vm (fixing issue #99)"

This reverts commit 3544acbcabcf598101fcb1f817394d86e505e513.

→ <<cset b8f8063895ca>>

[Jimmy Chu]

Will work on this after Apr 2nd.

Ideally, broker hostname is called broker01.oo.cnpaas.localhost. App domain will be apps.cnpaas.localhost, so we have a more consistent prod/staging/prod naming.

[LiAh Sheep]

But the problem is .localhost is used for 127.0.0.1 by default. I'm not sure if dnsmasq can set it properly?

[Jimmy Chu]

I believe they are configurable both on /etc/host and dnsmasq.

Set localhost to one IP, and abc.localhost to another IP.

[LiAh Sheep]

Found a workaround (finally). Short answer is to use the following in /etc/hosts:

::1 broker.openshift.local apps.openshift.local
::1 app01-test01.apps.openshift.local

192.168.33.10 broker.openshift.local apps.openshift.local
192.168.33.10 app01-test01.apps.openshift.local

Long answer:

Mac OS looks up .local domains by the following order:

1. IPV6 entries in /etc/hosts
2. IPV6 Bonjour service
3. IPV4 entries in /etc/hosts
4. IPV4 Bonjour service

The 5-second timeout occurs in Step 2, when Mac couldn’t find a machine matching the host name, then it proceeds to Step 3. If we add duplicate ::1 entries in /etc/hosts, we can trick Mac to think it is a IPV6 address and proceed to Step 3 directly.

Ref:

• http://stackoverflow.com/a/6992469/1316828 (answer by guns)