First of all, check this and ensure that you are deploying to the supported ubuntu version.
- Installation Problems
- Error: "You have not agreed to the Xcode license agreements"
- Error: checking whether the C compiler works... no
- Error: "fatal error: 'openssl/opensslv.h' file not found"
- Error: "TypeError: must be str, not bytes"
- Error: "ansible-playbook: command not found"
- Error: "Could not fetch URL ... TLSV1_ALERT_PROTOCOL_VERSION
- Bad owner or permissions on .ssh
- The region you want is not available
- AWS: SSH permission denied with an ECDSA key
- AWS: "Deploy the template" fails with CREATE_FAILED
- AWS: not authorized to perform: cloudformation:UpdateStack
- DigitalOcean: error tagging resource 'xxxxxxxx': param is missing or the value is empty: resources
- Windows: The value of parameter linuxConfiguration.ssh.publicKeys.keyData is invalid
- Docker: Failed to connect to the host via ssh
- Wireguard: Unable to find 'configs/...' in expected paths
- Ubuntu Error: "unable to write 'random state'" when generating CA password
- Connection Problems
- I'm blocked or get CAPTCHAs when I access certain websites
- I want to change the list of trusted Wifi networks on my Apple device
- Error: "The VPN Service payload could not be installed."
- Little Snitch is broken when connected to the VPN
- I can't get my router to connect to the Algo server
- I can't get Network Manager to connect to the Algo server
- Various websites appear to be offline through the VPN
- Clients appear stuck in a reconnection loop
- Wireguard: clients can connect on Wifi but not LTE
- "Error 809" or IKE_AUTH requests that never make it to the server
- Windows: Parameter is incorrect
- I have a problem not covered here
Look here if you have a problem running the installer to set up a new Algo server.
On macOS, you tried to install the dependencies with pip and encountered the following error:
Downloading cffi-1.9.1.tar.gz (407kB): 407kB downloaded
Running setup.py (path:/private/tmp/pip_build_root/cffi/setup.py) egg_info for package cffi
You have not agreed to the Xcode license agreements, please run 'xcodebuild -license' (for user-level acceptance) or 'sudo xcodebuild -license' (for system-wide acceptance) from within a Terminal window to review and agree to the Xcode license agreements.
No working compiler found, or bogus compiler options
passed to the compiler from Python's distutils module.
See the error messages above.
----------------------------------------
Cleaning up...
Command python setup.py egg_info failed with error code 1 in /private/tmp/pip_build_root/cffi
Storing debug log for failure in /Users/algore/Library/Logs/pip.log
The Xcode compiler is installed but requires you to accept its license agreement prior to using it. Run xcodebuild -license to agree and then retry installing the dependencies.
On macOS, you tried to install the dependencies with pip and encountered the following error:
Failed building wheel for pycrypto
Running setup.py clean for pycrypto
Failed to build pycrypto
...
copying lib/Crypto/Signature/PKCS1_v1_5.py -> build/lib.macosx-10.6-intel-2.7/Crypto/Signature
running build_ext
running build_configure
checking for gcc... gcc
checking whether the C compiler works... no
configure: error: in '/private/var/folders/3f/q33hl6_x6_nfyjg29fcl9qdr0000gp/T/pip-build-DB5VZp/pycrypto': configure: error: C compiler cannot create executables See config.log for more details
Traceback (most recent call last):
File "", line 1, in
...
cmd_obj.run()
File "/private/var/folders/3f/q33hl6_x6_nfyjg29fcl9qdr0000gp/T/pip-build-DB5VZp/pycrypto/setup.py", line 278, in run
raise RuntimeError("autoconf error")
RuntimeError: autoconf error
You don't have a working compiler installed. You should install the XCode compiler by opening your terminal and running xcode-select --install.
On macOS, you tried to install cryptography and encountered the following error:
build/temp.macosx-10.12-intel-2.7/_openssl.c:434:10: fatal error: 'openssl/opensslv.h' file not found
#include <openssl/opensslv.h>
^
1 error generated.
error: command 'cc' failed with exit status 1
----------------------------------------
Cleaning up...
Command /usr/bin/python -c "import setuptools, tokenize;__file__='/private/tmp/pip_build_root/cryptography/setup.py';exec(compile(getattr(tokenize, 'open', open)(__file__).read().replace('\r\n', '\n'), __file__, 'exec'))" install --record /tmp/pip-sREEE5-record/install-record.txt --single-version-externally-managed --compile failed with error code 1 in /private/tmp/pip_build_root/cryptography
Storing debug log for failure in /Users/algore/Library/Logs/pip.log
You are running an old version of pip that cannot download the binary cryptography dependency. Upgrade to a new version of pip by running sudo pip install -U pip.
You tried to install Algo and you see many repeated errors referencing TypeError, such as TypeError: '>=' not supported between instances of 'TypeError' and 'int' and TypeError: must be str, not bytes. For example:
TASK [Wait until SSH becomes ready...] *****************************************
An exception occurred during task execution. To see the full traceback, use -vvv. The error was: TypeError: must be str, not bytes
fatal: [localhost -> localhost]: FAILED! => {"changed": false, "failed": true, "module_stderr": "Traceback (most recent call last):\n File \"/var/folders/x_/nvr61v455qq98vp22k5r5vm40000gn/T/ansible_6sdjysth/ansible_module_wait_for.py\", line 538, in <module>\n main()\n File \"/var/folders/x_/nvr61v455qq98vp22k5r5vm40000gn/T/ansible_6sdjysth/ansible_module_wait_for.py\", line 483, in main\n data += response\nTypeError: must be str, not bytes\n", "module_stdout": "", "msg": "MODULE FAILURE"}
You may be trying to run Algo with Python3. Algo uses Ansible which has issues with Python3, although this situation is improving over time. Try running Algo with Python2 to fix this issue. Open your terminal and cd to the directory with Algo, then run: virtualenv -p `which python2.7` env && source env/bin/activate && pip install -r requirements.txt
You tried to install Algo and you see an error that reads "ansible-playbook: command not found."
You did not finish step 4 in the installation instructions, "Install Algo's remaining dependencies." Algo depends on Ansible, an automation framework, and this error indicates that you do not have Ansible installed. Ansible is installed by pip when you run python -m pip install -r requirements.txt. You must complete the installation instructions to run the Algo server deployment process.
You tried to install Algo and you received an error like this one:
Could not fetch URL https://pypi.python.org/simple/secretstorage/: There was a problem confirming the ssl certificate: [SSL: TLSV1_ALERT_PROTOCOL_VERSION] tlsv1 alert protocol version (_ssl.c:590) - skipping
Could not find a version that satisfies the requirement SecretStorage<3 (from -r requirements.txt (line 2)) (from versions: )
No matching distribution found for SecretStorage<3 (from -r requirements.txt (line 2))
It's time to upgrade your python.
brew upgrade python2
You can also download python 2.7.x from python.org.
You tried to run Algo and it quickly exits with an error about a bad owner or permissions:
fatal: [104.236.2.94]: UNREACHABLE! => {"changed": false, "msg": "Failed to connect to the host via ssh: Bad owner or permissions on /home/user/.ssh/config\r\n", "unreachable": true}
You need to reset the permissions on your .ssh directory. Run chmod 700 /home/user/.ssh and then chmod 600 /home/user/.ssh/config. You may need to repeat this for other files mentioned in the error message.
Algo downloads the regions from the supported cloud providers (other than Microsoft Azure) listed in the first menu using APIs. If the region you want isn't available, the cloud provider has probably taken it offline for some reason. You should investigate further with your cloud provider.
If there's a specific region you want to install to in Microsoft Azure that isn't available, you should file an issue, give us information about what region is missing, and we'll add it.
You tried to deploy Algo to AWS and you received an error like this one:
TASK [Copy the algo ssh key to the local ssh directory] ************************
ok: [localhost -> localhost]
PLAY [Configure the server and install required software] **********************
TASK [Check the system] ********************************************************
fatal: [X.X.X.X]: UNREACHABLE! => {"changed": false, "msg": "Failed to connect to the host via ssh: Warning: Permanently added 'X.X.X.X' (ECDSA) to the list of known hosts.\r\nPermission denied (publickey).\r\n", "unreachable": true}
You previously deployed Algo to a hosting provider other than AWS, and Algo created an ECDSA keypair at that time. You are now deploying to AWS which does not support ECDSA keys via their API. As a result, the deploy has failed.
In order to fix this issue, delete the algo.pem and algo.pem.pub keys from your configs directory and run the deploy again. If AWS is selected, Algo will now generate new RSA ssh keys which are compatible with the AWS API.
You tried to deploy Algo to AWS and you received an error like this one:
TASK [cloud-ec2 : Make a cloudformation template] ******************************
changed: [localhost]
TASK [cloud-ec2 : Deploy the template] *****************************************
fatal: [localhost]: FAILED! => {"changed": true, "events": ["StackEvent AWS::CloudFormation::Stack algopvpn1 ROLLBACK_COMPLETE", "StackEvent AWS::EC2::VPC VPC DELETE_COMPLETE", "StackEvent AWS::EC2::InternetGateway InternetGateway DELETE_COMPLETE", "StackEvent AWS::CloudFormation::Stack algopvpn1 ROLLBACK_IN_PROGRESS", "StackEvent AWS::EC2::VPC VPC CREATE_FAILED", "StackEvent AWS::EC2::VPC VPC CREATE_IN_PROGRESS", "StackEvent AWS::EC2::InternetGateway InternetGateway CREATE_FAILED", "StackEvent AWS::EC2::InternetGateway InternetGateway CREATE_IN_PROGRESS", "StackEvent AWS::CloudFormation::Stack algopvpn1 CREATE_IN_PROGRESS"], "failed": true, "output": "Problem with CREATE. Rollback complete", "stack_outputs": {}, "stack_resources": [{"last_updated_time": null, "logical_resource_id": "InternetGateway", "physical_resource_id": null, "resource_type": "AWS::EC2::InternetGateway", "status": "DELETE_COMPLETE", "status_reason": null}, {"last_updated_time": null, "logical_resource_id": "VPC", "physical_resource_id": null, "resource_type": "AWS::EC2::VPC", "status": "DELETE_COMPLETE", "status_reason": null}]}
Algo builds a Cloudformation template to deploy to AWS. You can find the entire contents of the Cloudformation template in configs/algo.yml. In order to troubleshoot this issue, login to the AWS console, go to the Cloudformation service, find the failed deployment, click the events tab, and find the corresponding "CREATE_FAILED" events. Note that all AWS resources created by Algo are tagged with Environment => Algo for easy identification.
In many cases, failed deployments are the result of service limits being reached, such as "CREATE_FAILED AWS::EC2::VPC VPC The maximum number of VPCs has been reached." In these cases, you must either delete the VPCs from previous deployments, or contact AWS support to increase the limits on your account.
You tried to deploy Algo to AWS and you received an error like this one:
TASK [cloud-ec2 : Deploy the template] *****************************************
fatal: [localhost]: FAILED! => {"changed": false, "failed": true, "msg": "User: arn:aws:iam::082851645362:user/algo is not authorized to perform: cloudformation:UpdateStack on resource: arn:aws:cloudformation:us-east-1:082851645362:stack/algo/*"}
This error indicates you already have Algo deployed to Cloudformation. Need to delete it first, then re-deploy.
You tried to deploy Algo to DigitalOcean and you received an error like this one:
TASK [cloud-digitalocean : Tag the droplet] ************************************
failed: [localhost] (item=staging) => {"failed": true, "item": "staging", "msg": "error tagging resource '73204383': param is missing or the value is empty: resources"}
failed: [localhost] (item=dbserver) => {"failed": true, "item": "dbserver", "msg": "error tagging resource '73204383': param is missing or the value is empty: resources"}
The error is caused because Digital Ocean changed its API to treat the tag argument as a string instead of a number.
- Download doctl
- Run
doctl auth init; it will ask you for your token which you can get (or generate) on the API tab at DigitalOcean - Once you are authorized on DO, you can run
doctl compute tag listto see the list of tags - Run
doctl compute tag delete environment:algo --forceto delete the environment:algo tag - Finally run
doctl compute tag listto make sure that the tag has been deleted - Run algo as directed
You tried to deploy Algo from Windows and you received an error like this one:
TASK [cloud-azure : Create an instance].
fatal: [localhost]: FAILED! => {"changed": false,
"msg": "Error creating or updating virtual machine AlgoVPN - Azure Error:
InvalidParameter\n
Message: The value of parameter linuxConfiguration.ssh.publicKeys.keyData is invalid.\n
Target: linuxConfiguration.ssh.publicKeys.keyData"}
This is related to the chmod issue inside /mnt directory which is NTFS. The fix is to place Algo outside of /mnt directory.
You tried to deploy Algo from Docker and you received an error like this one:
Failed to connect to the host via ssh:
Warning: Permanently added 'xxx.xxx.xxx.xxx' (ECDSA) to the list of known hosts.\r\n
Control socket connect(/root/.ansible/cp/6d9d22e981): Connection refused\r\n
Failed to connect to new control master\r\n
You need to add the following to the ansible.cfg in repo root:
[ssh_connection]
control_path_dir=/dev/shm/ansible_control_path
You tried to run Algo and you received an error like this one:
TASK [wireguard : Generate public keys] ********************************************************************************
[WARNING]: Unable to find 'configs/xxx.xxx.xxx.xxx/wireguard//private/dan' in expected paths.
fatal: [localhost]: FAILED! => {"msg": "An unhandled exception occurred while running the lookup plugin 'file'. Error was a <class 'ansible.errors.AnsibleError'>, original message: could not locate file in lookup: configs/xxx.xxx.xxx.xxx/wireguard//private/dan"}
This error is usually hit when using the local install option on a server that isn't Ubuntu 18.04. You should upgrade your server to Ubuntu 18.04. If this doesn't work, try removing *.lock files at /etc/wireguard/ as follows:
sudo rm -rf /etc/wireguard/*.lock
Then immediately re-run ./algo.
When running Algo, you received an error like this:
TASK [common : Generate password for the CA key] ***********************************************************************************************************************************************************
fatal: [xxx.xxx.xxx.xxx -> localhost]: FAILED! => {"changed": true, "cmd": "openssl rand -hex 16", "delta": "0:00:00.024776", "end": "2018-11-26 13:13:55.879921", "msg": "non-zero return code", "rc": 1, "start": "2018-11-26 13:13:55.855145", "stderr": "unable to write 'random state'", "stderr_lines": ["unable to write 'random state'"], "stdout": "xxxxxxxxxxxxxxxxxxx", "stdout_lines": ["xxxxxxxxxxxxxxxxxxx"]}
This happens when your user does not have ownership of the $HOME/.rnd file, which is a seed for randomization. To fix this issue, give your user ownership of the file with this command:
sudo chown $USER:$USER $HOME/.rnd
Now, run Algo again.
Look here if you deployed an Algo server but now have a problem connecting to it with a client.
This is normal.
When you deploy a Algo to a new cloud server, the address you are given may have been used before. In some cases, a malicious individual may have attacked others with that address and had it added to "IP reputation" feeds or simply a blacklist. In order to regain the trust for that address, you may be asked to enter CAPTCHAs to prove that you are a human, and not a Denial of Service (DoS) bot trying to attack others. This happens most frequently with Google. You can try entering the CAPTCHAs or you can try redeploying your Algo server to a new IP to resolve this issue.
In some cases, a website will block any visitors accessing their site through a cloud hosting provider due to previous, frequent DoS attacks originating from them. In these cases, there is not much you can do except deploy Algo to your own server or another IP that the website has not outright blocked.
This setting is enforced on your client device via the Apple profile you put on it. You can edit the profile with new settings, then load it on your device to change the settings. You can use the Apple Configurator to edit and resave the profile. Advanced users can edit the file directly in a text editor. Use the Configuration Profile Reference for information about the file format and other available options. If you're not comfortable editing the profile, you can also simply redeploy a new Algo server with different settings to receive a new auto-generated profile.
You tried to install the Apple profile on one of your devices and you received an error stating The "VPN Service" payload could not be installed. The VPN service could not be created. Client support for Algo VPN is limited to modern operating systems, e.g. macOS 10.11+, iOS 9+. Please upgrade your operating system and try again.
Little Snitch is not compatible with IPSEC VPNs due to a known bug in macOS and there is no solution. The Little Snitch "filter" does not get incoming packets from IPSEC VPNs and, therefore, cannot evaluate any rules over them. Their developers have filed a bug report with Apple but there has been no response. There is nothing they or Algo can do to resolve this problem on their own. You can read more about this problem in issue #134.
In order to connect to the Algo VPN server, your router must support IKEv2, ECC certificate-based authentication, and the cipher suite we use. See the ipsec.conf files we generate in the config folder for more information. Note that we do not officially support routers as clients for Algo VPN at this time, though patches and documentation for them are welcome (for example, see open issues for Ubiquiti and pfSense).
You're trying to connect Ubuntu or Debian to the Algo server through the Network Manager GUI but it's not working. Many versions of Ubuntu and some older versions of Debian bundle a broken version of Network Manager without support for modern standards or the strongSwan server. You must upgrade to Ubuntu 17.04 or Debian 9 Stretch, each of which contain the required minimum version of Network Manager.
This issue appears occasionally due to issues with MTU size. Different networks may require the MTU to be within a specific range to correctly pass traffic. We made an effort to set the MTU to the most conservative, most compatible size by default but problems may still occur.
If either your Internet service provider or your chosen cloud service provider use an MTU smaller than the normal value of 1500 you can use the reduce_mtu option in the file config.cfg to correspondingly reduce the size of the VPN tunnels created by Algo. Algo will attempt to automatically set reduce_mtu based on the MTU found on the server at the time of deployment, but it cannot detect if the MTU is smaller on the client side of the connection.
If you change reduce_mtu you'll need to deploy a new Algo VPN.
To determine the value for reduce_mtu you should examine the MTU on your Algo VPN server's primary network interface (see below). You might algo want to run tests using ping, both on a local client when not connected to the VPN and also on your Algo VPN server (see below). Then take the smallest MTU you find (local or server side), subtract it from 1500, and use that for reduce_mtu. An exception to this is if you find the smallest MTU is your local MTU at 1492, typical for PPPoE connections, then no MTU reduction should be necessary.
To check the MTU on your server, SSH in to it, run the command ifconfig, and look for the MTU of the main network interface. For example:
ens4: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1460
The MTU shown here is 1460 instead of 1500. Therefore set reduce_mtu: 40 in config.cfg. Algo should do this automatically.
When using ping you increase the payload size with the "Don't Fragment" option set until it fails. The largest payload size that works, plus the ping overhead of 28, is the MTU of the connection.
$ ping -4 -s 1432 -c 1 -M do github.com
PING github.com (192.30.253.112) 1432(1460) bytes of data.
1440 bytes from lb-192-30-253-112-iad.github.com (192.30.253.112): icmp_seq=1 ttl=53 time=13.1 ms
--- github.com ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 13.135/13.135/13.135/0.000 ms
$ ping -4 -s 1433 -c 1 -M do github.com
PING github.com (192.30.253.113) 1433(1461) bytes of data.
ping: local error: Message too long, mtu=1460
--- github.com ping statistics ---
1 packets transmitted, 0 received, +1 errors, 100% packet loss, time 0ms
In this example the largest payload size that works is 1432. The ping overhead is 28 so the MTU is 1432 + 28 = 1460, which is 40 lower than the normal MTU of 1500. Therefore set reduce_mtu: 40 in config.cfg.
$ ping -c 1 -D -s 1464 github.com
PING github.com (192.30.253.113): 1464 data bytes
1472 bytes from 192.30.253.113: icmp_seq=0 ttl=50 time=169.606 ms
--- github.com ping statistics ---
1 packets transmitted, 1 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 169.606/169.606/169.606/0.000 ms
$ ping -c 1 -D -s 1465 github.com
PING github.com (192.30.253.113): 1465 data bytes
--- github.com ping statistics ---
1 packets transmitted, 0 packets received, 100.0% packet loss
In this example the largest payload size that works is 1464. The ping overhead is 28 so the MTU is 1464 + 28 = 1492, which is typical for a PPPoE Internet connection and does not require an MTU adjustment. Therefore use the default of reduce_mtu: 0 in config.cfg.
If you don't wish to deploy a new Algo VPN (which is required to incorporate a change to reduce_mtu) you can change the client side MTU of WireGuard clients and Linux IPsec clients without needing to make changes to your Algo VPN.
For WireGuard on Linux, or macOS (when installed with brew), you can specify the MTU yourself in the client configuration file (typically wg0.conf). Refer to the documentation (see man wg-quick).
For WireGuard on iOS and Android you can change the MTU in the app.
For IPsec on Linux you can change the MTU of your network interface to match the required MTU. For example:
sudo ifconfig eth0 mtu 1440
To make the change take affect after a reboot, on Ubuntu 18.04 and later edit the relevant file in the /etc/netplan directory (see man netplan).
As of WireGuard for iOS 0.0.20190107 the default MTU is 1280, a conservative value intended to allow mobile devices to continue to work as they switch between different networks which might have smaller than normal MTUs. In order to use this default MTU review the configuration in the WireGuard app and remove any value for MTU that might have been added automatically by Algo.
If you're using 'Connect on Demand' on iOS and your client device appears stuck in a reconnection loop after switching from WiFi to LTE or vice versa, you may want to try disabling DoS protection in strongSwan.
The configuration value can be found in /etc/strongswan.d/charon.conf. After making the change you must reload or restart ipsec.
Example command:
sed -i -e 's/#*.dos_protection = yes/dos_protection = no/' /etc/strongswan.d/charon.conf && ipsec restart
Certain cloud providers (like AWS Lightsail) don't assign an IPv6 address to your server, but certain cellular carriers (e.g. T-Mobile in the United States, EE in the United Kingdom) operate an IPv6-only network. This somehow leads to the Wireguard app not being able to make a connection when transitioning to cell service. Go to the Wireguard app on the device when you're having problems with cell connectivity and select "Export log file" or similar option. If you see a long string of error messages like "Failed to send data packet write udp6 [::]:49727->[2607:7700:0:2a:0:1:354:40ae]:51820: sendto: no route to host then you might be having this problem.
Manually disconnecting and then reconnecting should restore your connection. To solve this, you need to either "force IPv4 connection" if available on your phone, or install an IPv4 APN, which might be available from your carrier tech support. T-mobile's is available for iOS here under "iOS IPv4/IPv6 fix", and here is a walkthrough for Android phones.
On Windows, this issue may manifest with an error message that says "The network connection between your computer and the VPN server could not be established because the remote server is not responding... This is Error 809." On other operating systems, you may try to debug the issue by capturing packets with tcpdump and notice that, while IKE_SA_INIT request and responses are exchanged between the client and server, IKE_AUTH requests never make it to the server.
It is possible that the IKE_AUTH payload is too big to fit in a single IP datagram, and so is fragmented. Many consumer routers and cable modems ship with a feature that blocks "fragmented IP packets." Try logging into your router and disabling any firewall settings related to blocking or dropping fragmented IP packets. For more information, see Issue #305.
TASK [cloud-digitalocean : Creating a droplet...] *******************************************
An exception occurred during task execution. To see the full traceback, use -vvv. The error was: NameError: name 'basestring' is not defined
fatal: [localhost]: FAILED! => {"changed": false, "msg": "name 'basestring' is not defined"}
If you get something like the above it's likely you're not using a python2 virtualenv.
Ensure running python2.7 drops you into a python 2 shell (it looks something like this)
user@homebook ~ $ python2.7
Python 2.7.10 (default, Feb 7 2017, 00:08:15)
[GCC 4.2.1 Compatible Apple LLVM 8.0.0 (clang-800.0.34)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>>
Then rerun the dependency installation explicitly using python 2.7
python2.7 -m virtualenv --python=`which python2.7` env && source env/bin/activate && python2.7 -m pip install -U pip && python2.7 -m pip install -r requirements.txt
The problem may happen if you recently moved to a new server, where you have Algo VPN.
-
Clear the Networking caches:
- Run CMD (click windows start menu, type 'cmd', right click on 'Command Prompt' and select "Run as Administrator").
- Type the commands below:
netsh int ip reset netsh int ipv6 reset netsh winsock reset -
Restart your computer
-
Reset Device Manager adaptors:
- Open Device Manager
- Find Network Adapters
- Uninstall WAN Miniport drivers (IKEv2, IP, IPv6, etc)
- Click Action > Scan for hardware changes
- The adapters you just uninstalled should come back
The VPN connection should work again
If you have an issue that you cannot solve with the guidance here, join our Gitter and ask for help. If you think you found a new issue in Algo, file an issue.