Unsolved
This post is more than 5 years old
5 Practitioner
•
274.2K Posts
0
2315
Step1 Docker install completes with no error's but no GUI
This is on a vSphere CentOS 7.1 minimal image. I let the script install docker. I opened ports at first but later decided to just turn the firewall off.
/data/vnest/vnest-main/configuration is completely empty, hence the nest errors. Which process is supposed to populate that directory? Where can we look for errors in that process?
mpeters2
24 Posts
0
July 7th, 2015 06:00
I am having the same issues with the portal access. It never appears on tcp/443 but always shows up on tcp/64443.
We are getting 2 IPs because one is from the physical interface and one is from the docker0 virtual interface that is configured automatically by Docker.
mpeters2
24 Posts
0
July 7th, 2015 08:00
Ok...so I got into the portal.
I pulled the latest code from GitHub and rebuilt it using the single node script.
At first the script was dying trying to get the hostname worked out from /etc/hostname. It refused to read the existing entry and would wipe the hostname file and try to rewrite it (but failed repeatedly). So I commented out the guts of the hosts_file_func function and made sure that my host information was present in both the /etc/hosts and /etc/hostname files.
The step1 script now completes as expected. The portal is still on tcp/64443 but at least the root/ChangeMe login now works properly.
Now on to licensing...
htarwala
11 Posts
1
July 7th, 2015 09:00
The second network interface was created by docker. See ifconfig output below:
We decided to start from scratch with reinstalling CentOS followed by ECS installation. Before we installed ECS we disabled IPv6, bumped up the VM memory from 30GB to 64GB and added an entry for the hostname to the hosts file. Not sure which one of the above fixed it but we can now access the UI on port 443.
The second interface interface named docker was recreated upon ECS installation.
Thanks,
Huzefa.
thakkd1
13 Posts
0
July 7th, 2015 10:00
Thanks for circling back. We have seen similar scenario (port 64443 vs 443) with not using the correct hostname. Thanks!
msalem_cfd6ea
19 Posts
0
July 7th, 2015 10:00
Correct docker create its interface but when you run command hostname -i it should not return as part of the result. I am not sure why it was return, Maybe Since I usually start by hostname and IP to the hosts file I did not run to this issue. All the changes you did fix the problem. With 30GB, it will be extremely slow. Adding the host file fixed the network issue.
BTW, did you try the latest version, it request the network interface and the hostname as parameters and the script use both value to figure out the ip address and add entry on hosts file.
Magdy
mpeters2
24 Posts
0
July 10th, 2015 07:00
I got the 443 access problem resolved by pulling the latest install scripts from GitHub and using them to do the build. One thing that was still not working in the script though was the hostname configuration.
On my CentOS 7 install, the script kept having a grep error (invalid range) when trying to examine the /etc/hostname file. The script would delete the existing file and try to write a new one with the system discovered hostname. That process consistently failed for me so I just commented out the function in the step1 script (as mentioned in my previous post).
I made sure the /etc/hostname file was present and correct and then the script ran perfectly with the required parameters (disks, ethadapter, and hostname). The services all start up (netstat -anp | more). Then once I disabled the firewall on the server (systemctl stop firewalld.service) I was able to reach the web interface on tcp/443 just fine. The default login also worked perfectly.
-Michael