Internet connection through a proxy / manual Ceph install possible?

[tuathan]

I connect to the internet through a proxy which (I don't have control of) and I've tried setting through the Proxmox web interface: "Datacenter" > "Options" the correct proxy as I have used elsewhere i.e. with format http://<username>:<password>@<proxy_ip_address>:3128/
but when I try to install Ceph via the Proxmox web interface which starts with "update available package list", it fails and the dialogue box does not even mention the proxy just "failed to fetch http://ftp.uk.debian..."

I'm wondering if:

1. The proxy setting is not saved or applied when set (I did also try adding apt.conf file with proxy details in /etc/apt and also doing an export http_proxy=... from CLI)
2. The Proxmox self signed cert is causing issue through the proxy (and how could this be resolved)

or

3. Can I use a work around and install Proxmox / Ceph (and anything else required for Proxmox) manually without an internet connection relatively easily? Is the manual Ceph install in Proxmox process documented somewhere online? Or even a list of packages which can be copied onto the Proxmox node and installed?

Thanks.

 

[t.lamprecht]

Hi,

the apt proxy configuration is also set to this proxy, but only after one run an update over the webinterface (Node → Updates) or over the CLI with pveupdate once after changing it. Do this on each node and it should work.

You can check if the apt proxy config got written out by looking at:

cat /etc/apt/apt.conf.d/76pveproxy

 

[tuathan]

Hi,

the apt proxy configuration is also set to this proxy, but only after one run an update over the webinterface (Node → Updates) or over the CLI with pveupdate once after changing it. Do this on each node and it should work.

You can check if the apt proxy config got written out by looking at:

cat /etc/apt/apt.conf.d/76pveproxy

Thanks for the info.

I reinstalled my 3 proxmox nodes (X.X.X.120,X.X.X.121,X.X.X,122 / with FQDNs etc.) and before creating any cluster tried to set the proxy and install ceph on each node.

On the first node (X.X.X.120) this worked without issue:

1. I accessed the web interface and set the proxy via "Datacenter" > "Options" as I indicated above (NB: No cluster at this point)
2. I then accessed the CLI and typed "pveupdate" once. This did return an error code but the file 76pveproxy was created.
3. "pveceph install" worked fine on this node no errors.
(I did not seem to need to edit the repositories for non-subscription -- as I'm currently investigating the use of Proxmox)

I followed the exact same process on the second node (X.X.X.121) but at step 3 I get errors:
-the repository is not signed
-407 Proxy Authorisation required [IP: <proxy ip address> <proxy port number>]

I have check on the second node and the file 76pveproxy was created correctly using step 1. and 2.

This also occured on node 3.

Am I missing something? Thanks

P.s. I couldn't see how to do Node>Updates from web interface.

 

[tuathan]

Thanks for the info.

I reinstalled my 3 proxmox nodes (X.X.X.120,X.X.X.121,X.X.X,122 / with FQDNs etc.) and before creating any cluster tried to set the proxy and install ceph on each node.

On the first node (X.X.X.120) this worked without issue:

1. I accessed the web interface and set the proxy via "Datacenter" > "Options" as I indicated above (NB: No cluster at this point)
2. I then accessed the CLI and typed "pveupdate" once. This did return an error code but the file 76pveproxy was created.
3. "pveceph install" worked fine on this node no errors.
(I did not seem to need to edit the repositories for non-subscription -- as I'm currently investigating the use of Proxmox)

I followed the exact same process on the second node (X.X.X.121) but at step 3 I get errors:
-the repository is not signed
-407 Proxy Authorisation required [IP: <proxy ip address> <proxy port number>]

I have check on the second node and the file 76pveproxy was created correctly using step 1. and 2.

This also occured on node 3.

Am I missing something? Thanks

P.s. I couldn't see how to do Node>Updates from web interface.

Update: It think it was an issue with the other nodes IP addresses i selected on my network. I managed to install Ceph on other two nodes using 3 steps above.

 

[t.lamprecht]

OK, good to hear.

 

[tuathan]

Just to finish this thread off. Above I was testing Proxmox Clustering using nested virtualization within Hyper-V. I installed Proxmox directly on a Dell R620 server node and followed above steps again (no issues) i.e.:

1. I accessed the web interface and set the proxy via "Datacenter" > "Options" as I indicated above (NB: No cluster at this point)
2. I then accessed the CLI and typed "pveupdate" once. This did return an error code (error 100) but the file /etc/apt/apt.conf.d/76pveproxy was created.
3. "pveceph install" worked fine on this node physical node: "installed ceph nautilus successfully"

 

[tuathan]

Sorry...this issue has persisted when installing Proxmox and trying to set the proxy for another two physical server nodes.

I follow above steps but at step two (pveupdate) the terminal quickly returns:

command 'apt-get update' failed: exit code 100 (NOTE: I also get this reponse when it works but after a much longer time has elapsed)

but /etc/apt/apt.conf.d/76pveproxy is correctly created but then if I do (pveceph install):

E. The repository .... is not signed
E. Failed to fetch ...
407 Proxy Authorization Required [IP: <proxy_ip_address> 3128]
.
.
.
apt failed during ceph installation (25600)

Which logs can I look at? Or how can I investigate?

I have managed to install CEPH on 1/3 physical server nodes so far. Is it possible to use that across the "Datacenter" (cluster) to enable (via the GUI) on the other nodes?

 

[tuathan]

On the server node where the proxy is working I can do:

curl -U <username>:<password> -x <proxy_ip_address>:3128 www.google.com

and get the google homepage. If i do this from another server node which was set up in exactly the same way (just has a different Ip address on my network) the above curl command returns a 407 Error.

 

[tuathan]

Well I dunno...tried again today on the other nodes where proxy authentication wasn't working. First I ran:

curl --proxy-ntlm --proxy-user %Username%:%Password% --proxy %PROXY_URL%:%PORT% %SITE_URL%

which worked!!

Then I could run pveupdate and pveceph install without issue.

Either I randomly happened to be connected to one of the enterprise NTLM proxy servers which my IT dept. has properly configured this afternoon or running the above curl command first provided the initial authentication needed for the subsequent pve commands...

 

[tuathan]

Well I dunno...tried again today on the other nodes where proxy authentication wasn't working. First I ran:

curl --proxy-ntlm --proxy-user %Username%:%Password% --proxy %PROXY_URL%:%PORT% %SITE_URL%

which worked!!

Then I could run pveupdate and pveceph install without issue.

Either I randomly happened to be connected to one of the enterprise NTLM proxy servers which my IT dept. has properly configured this afternoon or running the above curl command first provided the initial authentication needed for the subsequent pve commands...

Confirmed I've had to issue this curl command before pveupdate / pveceph install to properly authenticate on the proxy.

 

[t.lamprecht]

Hmm, a bit strange. Out of interest, what proxy software do you use?