Error: Cannot obtain a lock on data directory

[ekkis]

I'm trying to run bitcoind from the docker image like this:

docker run --rm -it --name bitcoind -p 8332 -v ${PWD}/bitcoind:/home/bitcoin/.bitcoin seegno/bitcoind:latest -server -printtoconsole

but I'm getting the error below:

/entrypoint.sh: assuming arguments for bitcoin
/entrypoint.sh: setting data directory to /home/bitcoin/.bitcoin

2016-07-03 05:57:12
[many empty lines]
2016-07-03 05:57:12 Bitcoin version v0.12.0 (2016-02-17 09:40:03 +0100)
2016-07-03 05:57:12 InitParameterInteraction: parameter interaction: -whitelistforcerelay=1 -> setting -whitelistrelay=1
2016-07-03 05:57:12 Error: Cannot obtain a lock on data directory /home/bitcoin/.bitcoin. Bitcoin Core is probably already running. No such file or directory.
Error: Cannot obtain a lock on data directory /home/bitcoin/.bitcoin. Bitcoin Core is probably already running. No such file or directory.
2016-07-03 05:57:12 Shutdown: In progress...
2016-07-03 05:57:12 StopNode()
2016-07-03 05:57:12 Shutdown: done

the directory I've created to host the blockchain (and presumably the wallet) looks like this:

drwxr-xr-x 2 ekkis staff 68 Jul 2 22:46 bitcoind/

so what am I doing wrong? the container doesn't have sufficient space in it for a sync so I need to write it to my local disk.

incidentally I'm running this on OSX using Boot2Docker, which I understand maps my /home folder automatically to the the docker-machine's file system. I've also verified that there is no bitcoind/.lock file.

[ekkis]

ok, I tried running it manually:

docker run --rm -it --name bitcoind -p 8332 -v ${PWD}/bitcoind:/home/bitcoin/.bitcoin --entrypoint=/bin/bash seegno/bitcoind:latest

and then, inside the container:

root@dbf45e2d4a42:/# bitcoind -server -printtoconsole -datadir=/home/bitcoin/.bitcoin
but now I see a different error:

2016-07-03 06:45:54 * Using 65.5MiB for in-memory UTXO set
2016-07-03 06:45:54 init message: Loading block index...
2016-07-03 06:45:54 Opening LevelDB in /home/bitcoin/.bitcoin/blocks/index
2016-07-03 06:45:54 IO error: /home/bitcoin/.bitcoin/blocks/index: Invalid argument
2016-07-03 06:45:54 : Error opening block database.

Do you want to rebuild the block database now?
: Error opening block database.

Do you want to rebuild the block database now?
2016-07-03 06:45:54 Aborted block database rebuild. Exiting.
2016-07-03 06:45:54 scheduler thread interrupt
2016-07-03 06:45:54 Shutdown: In progress...
2016-07-03 06:45:54 StopNode()
2016-07-03 06:45:54 Shutdown: done
root@dbf45

which I find a reference to here: boot2docker/boot2docker#794 so it seems there's a problem running bitcoind on VirtualBox (which I use for running docker on OSX)... however, curiously this is not the same error I get when I just run the container... a bit at a loss here...

[ruimarinho]

Indeed, this is related to how VirtualBox handles some file system calls. I honestly don't know the specifics but there's nothing we can do on the docker image side. Have you considered using Docker for Mac? It uses a custom file driver (osxfs) built specifically for Mac which does not present this issue.

[ekkis]

I appreciate your writing back. I removed Boot2Docker and installed the new (beta) Docker for Mac but I'm still unable to run. if I manually log into the container and run the server like this:

$ docker run --rm -it --name bitcoind -p 8332 -v ${PWD}/bitcoind:/home/bitcoin/.bitcoin --entrypoint=/bin/bash seegno/bitcoind:latest
root@8ecc1262ff23:/# bitcoind -server -printtoconsole -datadir=/home/bitcoin/.bitcoin

it seems to work but running it from the host:

$ docker run --rm -it --name bitcoind -p 8332 -v ${PWD}/bitcoind:/home/bitcoin/.bitcoin seegno/bitcoind:latest -server -printtoconsole

fails with:

chown: changing ownership of '/home/bitcoin/.bitcoin/.cookie': Operation not permitted
chown: changing ownership of '/home/bitcoin/.bitcoin/.lock': Operation not permitted
chown: changing ownership of '/home/bitcoin/.bitcoin': Operation not permitted

can you spot what the issue might be? here's the directory I'm mounting:

$ls -alF bitcoind

drwx------ 4 ekkis staff 136 Jul 5 11:38 bitcoind/

[ekkis]

I also had a look at what the original entry point does and tried running with an alternative directory but it too failed:

$ docker run --rm -it --name bitcoind -p 8332 -e BITCOIN_DATA=/data -v ${PWD}/bitcoind:/data seegno/bitcoind:latest -server -printtoconsole

/entrypoint.sh: assuming arguments for bitcoind
chown: changing ownership of '/data/.cookie': Operation not permitted
chown: changing ownership of '/data/.lock': Operation not permitted
chown: changing ownership of '/data': Operation not permitted

[ekkis]

if I log into the container and run the entry point (since I replaced it in my invocation to the container), I get an error:

root@811e3f6c65b1:/# ./entrypoint.sh

./entrypoint.sh: 4: [: =: unexpected operator
./entrypoint.sh: 10: [: =: unexpected operator

and curiously this fails too:

root@811e3f6c65b1:/# /bin/sh
# if [ $(echo "$1" |cut -c1) = "-"]; then
> echo "ok"
> fi

/bin/sh: 1: [: missing ]

I don't get it

[ekkis]

so the problem seems to be that if $1 is empty it needs to be written like this:

if [ "$(echo $1 |cut -c1)" = "-"]; then

but I see that entrypoint.sh is called without parameters in the Dockerfile, so I'm wondering why I see the message:

/entrypoint.sh: assuming arguments for bitcoind

in my penultimate message...

[ekkis]

ah... I get it now. CMD ["bitcoind"] is actually a parameter call to the entry point, and now if I log into the container and run /entrypoint.sh "bitcoind" I get the error messages failing to change the ownership. now the question is why?

I presume the container runs under the docker host id, which wouldn't have permission to change ownership of my folder... so the question is: how do I enable the docker host to own it?

[ekkis]

I've created a post on the docker site for this: https://forums.docker.com/t/failure-to-chown-a-directory-on-the-host/17445

...but in discussing this with some of the blokes on irc#docker it was suggested that the chown in the entry point should be removed...

[ekkis]

final note: I killed all containers running, removed all volumes, all images, and ran the command again. it worked as it should. so I guess it was some bad juju from the times I was running with boot2docker. so we're all good now. thanks a bunch!

[ruimarinho]

Sorry for the lack of feedback here @ekkis, but I'm glad you're on track again!

[ekkis]

yes, thank you for putting all this together. your efforts to maintain this make it child's play to put things together and saves those of us playing with this stuff a ton of work. cheers!