Use rsync with gitlab-ci.yml to sync Wordpress theme folders

Question

While trying to use .gitlab-ci.yml to automatically deploy my theme to my production server, I receive a error:

The command executed: using a ubuntu:16.04 image

$ apt update -y && apt install openssh-client sshpass rsync -y
$ rsync -avh --progress --delete --rsh="sshpass -p $STAGE_FTP_PASS ssh -o StrictHostKeyChecking=no " --exclude=.git ./ $STAGE_FTP_USER@my.wp.address/project-folder/wp-content/themes/sg-coesfeld-theme/

The error result:

sshpass: invalid option -- 'o'
protocol version mismatch -- is your shell clean?
(see the rsync man page for an explanation)
rsync error: protocol incompatibility (code 2) at compat.c(176) [sender=3.1.1]
ERROR: Job failed: exit code 1

I am using the exact same .gitlab-ci.yml , server, FTP account, runner and gitlab hosting successfully to auto-deploy in other git repos.

The only thing that has changed is the source git repository and the /project-folder/ .

To be honest I do not know where to start and how to track down this issue, so I am happy about any suggestions.

Answer 1

Your $STAGE_FTP_PASS must be empty so the command would become

sshpass -p ssh -o StrictHostKeyChecking=no 

so sshpass would think -p ssh is the password and -o StrictHostKeyChecking=no is the command.

Answer 2

As @pynexj already has said, $STAGE_FTP_PASS was an empty string, which caused the error. I have tried this using echo $STAGE_FTP_USER inside my script.

While inserting the secret variables, I have checked the box [X] Protected , which has an explanation attached: This variable will be passed only to pipelines running on protected branches and tags.

Searching for protected branches I have noticed, that in Settings > Repository > Protected Branches no protected branch was set.

So I could choose between unchecking the [ ] Protected checkbox, so that every branch can use the secret variables, or go with the more secure approach and add my branch (for example master ) as a protected branch.

Now everything works great again. Thanks again to @pynexj for pointing out what causes the problem in the script.