[Solved] Abuse cURL to communicate with Redis

I want to send a PING to Redis to check if the connection is working, now I could just install redis-cli, but I don’t want to and curl is already there. So how can I abuse curl to do that? Basically I need to turn off what’s send here:

> GET / HTTP/1.1
> User-Agent: curl/7.22.0 (x86_64-pc-linux-gnu) libcurl/7.22.0 OpenSSL/1.0.1 zlib/1.2.3.4 libidn/1.23 librtmp/2.3
> Host: localhost:6379
> Accept: */*
> 
-ERR wrong number of arguments for 'get' command
-ERR unknown command 'User-Agent:'
-ERR unknown command 'Host:'
-ERR unknown command 'Accept:'

I was able to get rid of the User-Agent altogether by adding -A "", but I can’t find anything else for the rest. Any idea how I can do that?

Enquirer: Mahoni

||

Solution #1:

When you want to use curl, you need REST over RESP, like webdis, tinywebdis or turbowebdis. See https://github.com/markuman/tinywebdis#turbowebdis-tinywebdis–cherrywebdis

$ curl -w '
' http://127.0.0.1:8888/ping
{"ping":"PONG"}

Without a REST interface for redis, you can use netcat for example.

$ (printf "PING
";) | nc <redis-host> 6379 
+PONG

For password protected redis you can use netcat like this:

$ (printf "AUTH <password>
";) | nc <redis-host> 6379
+PONG

With netcat you have to build the RESP protocol by your self. See http://redis.io/topics/protocol

update 2018-01-09

I’ve build a powerfull bash function which pings the redis instance at any cost over tcp

    function redis-ping() {
            # ping a redis server at any cost
            redis-cli -h $1 ping 2>/dev/null || 
                    echo $((printf "PING
";) | nc $1 6379 2>/dev/null || 
                    exec 3<>/dev/tcp/$1/6379 && echo -e "PING
" >&3 && head -c 7 <&3)
    }

usage redis-ping localhost

Respondent: Markus

Solution #2:

Not curl, but doesn’t require a HTTP interface or nc (great for something like a container where you don’t have nc installed)

exec 3<>/dev/tcp/127.0.0.1/6379 && echo -e "PING
" >&3 && head -c 7 <&3

Should give you

+PONG

You can read more about what’s going on from this fantastic article.

Respondent: Joel B

Solution #3:

I needed to add a sleep to the nc provided by @Markus to get it to work from a remote system:

(printf "PING
"; sleep 1) | nc remote.redis.hostname 6379

See Request/Response protocols and RTT: Redis Pipelining for details.

Respondent: Peter M

Solution #4:

You can also use telnet localhost 6379 and if connection will be successful, type ping

To go out use quit

Respondent: Dennis Meissel

Solution #5:

I had docker running on mac (Catalina).
I ran a redis container using

docker run --name redis-mac -p 6379:6379 -d redis

Then I tested the connection using the computer’s IP

echo PING | nc 192.168.1.100 6379

Recieved

+PONG

Using the IP from docker inspect redis-mac didn’t work.

Actually I was testing Spring boot application having redis as cache provider. So I needed to test the connection .
If your’e new to docker, here’s a bit explanation of the first command :
Run redis as a container , with host port 6379 and docker port 6379 , and name the container as redis-mac . Now docker daemon will create a container if the image is already available , else it will pull the image from docker hub and then run the container.
You can do docker ps to check if the container is running.

Respondent: App Work

Solution #6:

Elaborating on @Joel B’s great answer. I needed this in a shell script in docker container, with no curl, no redis-cli and no nc…
The REDIS I’m testing is the public redis-ha Helm chart for kubernetes from here: https://github.com/helm/charts/tree/master/stable/redis-ha

Variables to set are:

  • REDIS_HOST = host name (DNS address or IP) containing the redis master and sentinel
    (break this into separate hosts if they’re separate, and change the ports
    if you need to – but in the helm chart the master/slave and sentinel are
    in the same pod and use the standard ports)

  • REDIS_STARTUP_TIMEOUT = maximum seconds to wait before giving up – default to 10 mins

  • REDIS_STARTUP_RETRY = seconds to wait between tests – default to 15 seconds

  • DEBUG = set this to true to echo the failed responses

The intricacies of the technique are described in the comments (I suggest you keep the comments in place to save your colleagues – and future self – from certain madness when trying to decipher the angle brackets)

# wait for 10 mins and check again every 15 seconds
let n=${REDIS_STARTUP_TIMEOUT:-600}
let m=${REDIS_STARTUP_RETRY:-15}
ready=false
while ((n > 0)); do
    # These scripts are the best way to check if redis is running without having access to nc, curl or redis-cli
    # They write a "PING" to the redis and sentinel ports on the hostname "dc-ecm-redis-ha"
    # and look for a "PONG+" in return.
    #
    # Detailed explanation:
    # -  3<>/dev/tcp... opens a file handle identified as #3 for input and output on the tcp host and port
    #    The host $REDIS_HOST is defined in DNS by the Kubernetes _service_, and the port is for redis or sentinel
    #    (Uses linux's low-level network-as-filesystem support. Think of it as a sort of poor-man's telnet)
    # -  "PING" followed by carriage-return is sent to the port by redirecting to the handle with >&3
    # -  The response from the port is sent to the head command with <&3
    # -  The first 5 bytes of the response are captured with -c 5. This removes the linebreak (
) from the response
    # -  Standard shell $() around the whole thing stores the result in a variable (master or sentinel)
    # -  When Redis is NOT ready, the result is generally a failure in the exec or some other error, which goes
    #    to stderr, so we wrap it in  { } > 2>&1 to capture that error in the variable too.
    # -  Success is measured by "+PONG" being in the variable
    # -  If you set the variable DEBUG to "true" (or "TRUE" -> the {,,} lower-cases it) the failed responses are echoed
    # -  (There are easier ways to do this if you have redis-cli installed or nc, but typically you don't on a docker container)
    # -  The whole thing waits n seconds for Redis to be ready, checking every m seconds
    #
    master=$( { exec 3<>/dev/tcp/${REDIS_HOST}/6379 && echo -e "PING
" >&3 && head -c 5 <&3; } 2>&1 )
    sentinel=$( { exec 3<>/dev/tcp/${REDIS_HOST}/26379 && echo -e "PING
" >&3 && head -c 5 <&3; } 2>&1 )
    if [ "$sentinel" = "+PONG" -a "$master" = "+PONG" ]; then ready=true;
       break;
   else echo "$(date) : Waiting $n more seconds for Redis master and sentinel to respond to PING"
        [[ "${DEBUG,,}" = "true" ]] && echo "master response was [$master]";
        [[ "${DEBUG,,}" = "true" ]] && echo "sentinel response was [$sentinel]";
        sleep $m
        ((n-=m))
    fi
done

if [ "$ready" = true ]
    then echo "$(date) : REDIS is ready"
    # do something when Redis is ready
else
    echo "$(date) : ERROR: REDIS is still not ready. Giving up waiting"
    # do something when Redis fails
fi
Respondent: Rhubarb

Solution #7:

In case Redis instance requires pass this might help:

$ exec 3<>/dev/tcp/127.0.0.1/6379 && echo -e "AUTH {requirepass-string-from-redis-config} 
 PING
" >&3 && head -c 12 <&3
+OK
+PONG

Replace {requirepass-string-from-redis-config} with string from ‘requirepass’ redis.conf

Respondent: Igorium

The answers/resolutions are collected from stackoverflow, are licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0 .

Leave a Reply

Your email address will not be published.