Bash Script fails with error: OCI runtime exec failed

Question

I am running the below script and getting error.

#!/bin/bash
webproxy=$(sudo docker ps -a --format "{{.Names}}"|grep webproxy)
webproxycheck="curl -k -s https://localhost:\${nginx_https_port}/HealthCheckService"
if [ -n "$webproxy" ] ; then
  sudo docker exec $webproxy sh -c "$webproxycheck"
fi

Here is my docker ps -a output

$sudo docker ps -a --format "{{.Names}}"|grep webproxy
webproxy-dev-01
webproxy-dev2-01

when i run the command individually it works. For Example:

$sudo docker exec webproxy-dev-01 sh -c 'curl -k -s https://localhost:${nginx_https_port}/HealthCheckService'
HEALTHCHECK_OK

$sudo docker exec webproxy-dev2-01 sh -c 'curl -k -s https://localhost:${nginx_https_port}/HealthCheckService'
HEALTHCHECK_OK

Here is the error i get.

$ sh healthcheck.sh
OCI runtime exec failed: exec failed: container_linux.go:348: starting container process caused "exec: \"webproxy-dev-01\": executable file not found in $PATH": unknown

Could someone please help me with the error. Any help will be greatly appreciated.

You don't need `sudo` or `docker exec` for this at all; you should be able to just invoke that `curl` command against a published port for your container. — David Maze, Aug 29 '20 at 17:46
`sudo docker` is necessary on some platforms like Red Hat, but e.g. on Debian-based platforms you should simply add your user to the `docker` group and then run `docker` without `sudo`. — tripleee, Oct 10 '20 at 09:06

tripleee · Accepted Answer · 2020-10-10T09:04:45.700

Because the variable contains two tokens (on two separate lines) that's what the variable expands to. You are running

sudo docker exec webproxy-dev-01 webproxy-dev2-01 ...

which of course is an error.

It's not clear what you actually expect to happen, but if you want to loop over those values, that's

for host in $webproxy; do
    sudo docker exec "$host" sh -c "$webproxycheck"
done

which will conveniently loop zero times if the variable is empty.

If you just want one value, maybe add head -n 1 to the pipe, or pass a more specific regular expression to grep so it only matches one container. (If you have control over these containers, probably run them with --name so you can unambiguously identify them.)

score 0 · Answer 2 · edited Sep 01 '20 at 20:18

Based on your given script, you are trying to "exec" the following

sudo docker exec webproxy-dev2-01
webproxy-dev-01 sh -c "curl -k -s https://localhost:${nginx_https_port}/HealthCheckService"

As you see, here is your error.

sudo docker exec webproxy-dev2-01
webproxy-dev-01 [...]

The problem is this line:

webproxy=$(sudo docker ps -a --format "{{.Names}}"|grep webproxy)

which results in the following (you also posted this):

webproxy-dev2-01 
webproxy-dev-01

Now, the issue is, that your docker exec command now takes both images names (coming from the variable assignment $webproxy), interpreting the second entry (which is webproxy-dev-01 and sepetrated by \n) as the exec command. This is now intperreted as the given command which is not valid and cannot been found: That's what the error tells you.

A workaround would be the following:

webproxy=$(sudo docker ps -a --format "{{.Names}}"| grep webproxy | head -n 1)

It only graps the first entry of your output. You can of course adapt this to do this in a loop.

A small snippet:

#!/bin/bash

webproxy=$(sudo docker ps -a --format "{{.Names}}"| grep webproxy )
echo ${webproxy}

webproxycheck="curl -k -s https://localhost:\${nginx_https_port}/HealthCheckService"

while IFS= read -r line; do
    if [ -n "$line" ] ; then
        echo "sudo docker exec ${line} sh -c \"${webproxycheck}\""
    fi   
done <<< "$webproxy"

In the OP's original code, the variable is not quoted, so the newline between the values is normalized to a single space by the shell. — tripleee, Aug 29 '20 at 14:16

Bash Script fails with error: OCI runtime exec failed

2 Answers2