Git Remote: Error: fatal: protocol error: bad line length character: Unab

Question

I set up a git server and want now to push initially my repo from the client. I used git push origin master and get this error message:

fatal: protocol error: bad line length character: Unab

I don't know what's wrong. I don't know what "Unab" is. I tried to resize the shell but it is still "Unab". I cannot find a solution for this error message.

I setup the server with "authorized_keys" and SSH. (I can connect to it, using SSH.)

It seems to be a git problem?

BTW: The server is set up in a Windows 7 VM

Answer 1

This error message is a bit obtuse, but what it's actually trying to tell you is that the remote server didn't reply with a proper git response. Ultimately, there was a problem on the server running the git-receive-pack process.

In the Git protocol, the first four bytes should be the line length. Instead, they were the characters Unab... which is probably the beginning an error message of some kind. (ie, it's probably "Unable to..." do something).

What happens when you run ssh <host> git-receive-pack <path-to-git-repository>? You should see the error message that your git client is barfing on and you may be able to correct it.

Answer 2

I had similar issue, but the exact error message was:

fatal: protocol error: bad line length character: Usin

This is in Windows, with GIT_SSH set to the path of plink.exe of PuTTY.

Possible problems and solutions:

• Make sure the path to plink.exe is correct. Unix style path works fine too, for example /c/work/tools/PuTTY/plink.exe
• Make sure the key agent of PuTTY (pageant.exe) is running
• Make sure the key agent contains a valid key to access the server

Answer 3

For GitExtension users:

I faced the same issue after upgrading git to 2.19.0

Solution:

Tools > Settings > Git Extensions > SSH

Select [OpenSSH] instead of [PuTTY]

Answer 4

I had the same kind of problem after installing GIT on Windows. At first it worked; then, a day later (after a PC reboot), it didn't anymore, and I got this:

$ git pull
fatal: protocol error: bad line length character: git@

The problem was that after the reboot, the automatically started Putty "pageant.exe" didn't have the private key active anymore. When you add a key in pageant, it's not a persistent setting by default. I just had to add the key again, and it worked fine. So, for that case, it's necessary to make pagenant load the key automatically, as discussed here:

https://www.digitalocean.com/community/tutorials/how-to-use-pageant-to-streamline-ssh-key-authentication-with-putty

Answer 5

Maybe you have a statement in the server's .bashrc that produces output. I, for example had this:

[[ -s "$HOME/.rvm/scripts/rvm" ]] && source "$HOME/.rvm/scripts/rvm"
rvm use ruby-1.9.3-p194@rails32

In this case the output from the rvm use will be (wrongly) interpreted as coming from git. So replace it by:

rvm use ruby-1.9.3-p194@rails32 > /dev/null

Answer 6

After loading the SSH private key in Git Extensions, this issue gets resolved.

Answer 7

You can redirect any output from .bashrc to stderr:

# inside .bashrc
echo 'some error/warning/remind message' 1>&2

git will ignore this symbols

Answer 8

I had a similar problem on Windows using Git Bash. I kept getting this error when trying to do a git clone. The repository was on a Linux box with GitLab installed.

git clone git@servername:path/to/repo
fatal: protocol error: bad line length character: git@

I made sure the ssh key was generated. The public key was added on GitLab. The ssh-agent was running and the generated key was added (github link).

I ran out of options and then finally tried closing Git Bash and opening it again by right clicking 'Run as Administrator'. Worked after that.

Answer 9

For me it was becuase I recently added

RequestTTY force

into .ssh/config

commenting this out allowed it to work

Answer 10

This might help someone. When I was trying to clone a project from an EC2 instance, I was getting the below error:

Cloning into 'repo1'...
fatal: protocol error: bad line length character: logi

The resolution for me includes the below steps:

1. Ensure SSH key (public) is added/updated in the EC2 instance.
2. Ensure authentication agent (in my case its Pageant=Putty Authentication Agent) is running and the corresponding private key is loaded.

3. Use the EC2 SSH key ID for the public key for git clone. Example:

   git clone ssh://{SSH Key ID}@someaccount.amazonaws.com/v1/repos/repo1

Answer 11

Check your startup files on the account used to connect to the remote machine for "echo" statements. For the Bash shell these would be your .bashrc and .bash_profile etc. Edward Thomson is correct in his answer but a specific issue that I have experienced is when there is some boiler-plate printout upon login to a server via ssh. Git will get the first four bytes of that boiler-plate and raise this error. Now in this specific case I'm going to guess that "Unab" is actually the work "Unable..." which probably indicates that there is something else wrong on the Git host.

Answer 12

In my case after fetch it was written: fatal: protocol error: bad line length character: Pass. Also after push I got: fatal: protocol error: bad line length character: git@ Done.

After reboot of Windows I had to start "PuTTY agent" (pageant.exe) again and add a private key that disappeared from a list of keys.

Answer 13

If you use Putty. Then make sure to have Pageant running and your private key is loaded in Pageant (mouse right-click on Pageant icon on the Taskbar and click "View keys" on the menu that pops up).

Otherwise when you do in cmd.exe :

git clone ssh://name@host:/path/to/git/repo.git

you get this message "fatal: protocol error: bad line length character:"

Answer 14

i also encounter that error once in a while, but when it does, it means that my branch is not up-to-date so i have to do git pull origin <current_branch>

Answer 15

FYI I got this same error message after I upgraded a CentOS6 container to CentOS7 -- some git operations started failing when building the container, e.g.

# git remote show origin
fatal: protocol error: bad line length character: Inva

Running ssh gave me an error I could search on:

# ssh git@bitbucket.org
Invalid clock_id for clock_gettime: 7

That led me to https://github.com/wolfcw/libfaketime/issues/63 where I realized I had forgotten I had a LD_PRELOAD=/usr/local/lib/faketime/libfaketime.so.1 in a parent Dockerfile. Commenting that out fixed the error.

Answer 16

In my case the problem was 32-bit Putty and pageant.exe - it can't communicate with 64-bit TortoisePlink.exe. Replacing 32-bit Putty with a 64-bit version solved the problem.

Answer 17

I had the same error "fatal: protocol error: bad line length character: shmi" Where the shmi is user name in my case. I switched SSH from PuTTY to OpenSSH in "Git Extensions->Settings->SSH". It helped.

Answer 18

I had the same problem as Christer Fernstrom. In my case it was a message I had put in my .bashrc that reminds me do do a backup when I haven't done one in a couple of days.

Answer 19

The following may help someone: When trying to clone a project I have on my AWS EC2 instance I was getting the following error:

Cloning into 'AWSbareRepo'...
fatal: protocol error: bad line length character: Plea

This was caused by trying to ssh as root instead of EC2-USER. if you actually ssh without doing a git clone... you will see the error msg in something along the lines of "Please login with ec2-user" Once I did a git clone as a ec2-user it was good.

Answer 20

Git doesn't prompt for password and fails with similar cryptic message "fatal: protocol error: bad line length character: user" if you don't have your private key authentication setup as well.

https://www.digitalocean.com/community/tutorials/how-to-configure-ssh-key-based-authentication-on-a-linux-server tells how to specify public key on the server. Basically add the public key to ~/.ssh/authorized_keys or ~/.ssh/authorized_keys2

I had to struggle a bit on how to provide private key to the Git Bash on the windows machine. Dan McClain's answer in https://serverfault.com/questions/194567/how-do-i-tell-git-for-windows-where-to-find-my-private-rsa-key/382801#382801 describes that. One addition to his answer, in my case the private key file was expected to be named id_rsa.pub

Answer 21

For me adding the same host details into Putty with the private key (convert with puttygen) worked. Any git bash commands after that had no problems.

Answer 22

TL;DR: Do not omit username@ in your remote URLs when on Windows.

On Linux and on Windows with the default ssh, you can omit the username from remote URLs, like so:

git clone server-name:/srv/git/repo-name

Because ssh's default behavior is to just use whatever username you're currently logged in with. If you're on Windows and have set up git to use plink.exe so that you can use the key loaded in your pageant, then this will not work, because plink does not have this same automatic username behavior, resulting in those cryptic error messages, because it will prompt for the username:

$ plink server-name
login as: _

Versus:

$ plink username@server-name
...logs you in...

If you already cloned a repository somehow you can fix the remotes in your .git/config by adding the username@ to the remote URL.

Answer 23

Check if Shell access is allowed on the server.

Answer 24

The error transformed in: fatal: protocol error: bad line length character: fata

after adding the location of git-upload-pack to the system path.

The problem seem to be an apostrophe added around the repository name: Looking with a tool like Process Monitor (from sys internals), that were added by the git client. It seems to be a git specific windows problem.

I tried the same command line in the prompt of the server: the full error was "fatal: not a given repository (or any of the parent directories): .git"

In conclusion, to me it seems like a software bug. Be advised that I am not a git expert, it is first time I use git, i come from subversion and perforce.

Answer 25

We ran into this as well.

Counting objects: 85, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (38/38), done.
Writing objects: 100% (38/38), 3.38 KiB | 0 bytes/s, done.
Total 38 (delta 33), reused 0 (delta 0)
Auto packing the repository for optimum performance.
fatal: protocol error: bad line length character: Remo
error: error in sideband demultiplexer

I don't know the gitty details about what went wrong, but in our case what triggered it was that the disk on the server was full.

Answer 26

It could be a security access on your machine, are you running Pageant (which is a putty agent)?

Answer 27

you can always have http link to your git project. You can use that instead of ssh link. This is just an option you have

Answer 28

Well, I had this same issue (Windows 7). Try to get repo by password. I use Git Bash + Plink (environment variable GIT_SSH) + Pageant. Deleting GIT_SSH (temporary) helps me. I don't know why I can't use login by pass and login with RSA at the same time...

Answer 29

Late answer here, but hope it will help someone. If its a protocol error, it has to do something with your local git not able to communicate to the remote git. This can happen if you cloned the repo via ssh and sometime later, you lost the keys to the repo or your ssh agent cannot find those keys anymore.

Solution

1. Generate a new key and add it your git repo or configure your ssh agent to load the keys if you still have the keys with you & not with someone else ;)

2. Another quick fix is to go to your .git directory and edit the config file's [remote "origin"] url from git to http so that ssh keys are not needed to push and it will revert to asking your username and password.

   [remote "origin"]
   url = git@gitlab.*****.com:****/****.git
   fetch = +refs/heads/*:refs/remotes/origin/*
   

Change to

    [remote "origin"]
    url = http://gitlab.*****.com/****/****.git
    fetch = +refs/heads/*:refs/remotes/origin/*

Answer 30

Changing the ssh exectuable from builtin to nativ under settings/version control/git did the trick for me.