I am deploying an ear file inside a docker container. Ear contains a properties file which tells where the logs need to be stored. But inside the container, the directory doesn't exist. So I am wondering where the logs are getting written, if they are getting written at all ? And if not, then how do I specify where the logs should be written inside container ?
Related
I wish I could update the settings of the application by changing the local profile.
I use "volume" to bind a local directory, for example:
docker run -v D:\test:/app
But when the container is running, all files in /app are emptied, because D:\test does not have any files.
Is there any way I can achieve my goal
Your question is a bit unclear. I guess you're problem is the following: You want to bind mount your app directory, but it is initially empty and will stay empty since the bind mount overwrites everything put into /app during build
I usually use two different ways:
Put your profile into host directory D:\test (if applicable). This is also a viable strategy for e.g. the source code of nodejs apps
During build, put your profile into /app_temp. Then create an entry point which moves /app_temp into /app. If you want to persist the profile through multiple build/run phases, it has to be inside the build context (which is likely not D:\test) on your host.
You need to change a bit the way your application is organized. put all the settings in their own directory and have the application read them from there. Then you can map only the settings folder you the host one.
Another option is to map the host folder to a temporary folder inside the container and have the ENTRYPOINT script update your files (by copying them over) and then run your application.
Docker was not meant to be used for the workflow you are trying to setup and for this reason you need to do some extra work.
because D:\test does not have any files.
That the way it works. Volume type you use is bind mount, i.e. you mount file system, using mount point mapped to a host directory.
According to documentation:
With bind mounts, we control the exact mountpoint on the host. We can use this to persist data, but it’s often used to provide additional data into containers.
You have two options here (both imply host data should exist in advance):
Bind to a folder, containing configuration data. As you showed.
It is possible to bind only file:
docker run -v D:\test\config.json:/app/config.json
While binding to a file, if it does not exist beforehand, docker-daemon would think it is a directory and will create directory, both in container and on the host.
you mount file system, using mount point mapped to a host directory
Hence, if host directory is empty mounted file system would also be empty.
I have an image based on opencpu/base. It starts an apache based server, and then invokes R scripts everytime sombody calls an API endpoint.
One of those scripts tries to write a file to a location in the container. When I mount a folder into that location, it works on my Windows machine, but not on Ubuntu.
I've tried using named volumes on Ubuntu, but it does not work either. When I run bash inside the container interactively on Ubuntu, I can write and read the mounted volume just fine. But the apache process cannot.
Does anybody have some hints what could be going on here?
When you log in interactively to the container, you will have root permissions.
Apache usually runs as another user (www-data), and that user must have read permissions on the folder that you want it to read.
Make sure that the permissions of the folder matches the user that will read it.
I have a WebLogic 12.1.2 image that I'm looking to auto deploy an application ear on to.
I can do this when building the image using the Dockerfile i.e.
COPY some.ear /u01/domains/mydomain/autodeploy
The container run using this image runs fine and I can reach my application REST interface with Postman.
However, if I restart that container, the autodeployed application no longer exists in the container's WebLogic console and hence Postman calls fail.
The ear file is still in the autodeploy directory on restart.
Has anyone seen this behaviour before?
I've found a little hack around this.
My container has the WebLogic autodeploy directory mounted to a host directory containing the ear. If I rename the ear e.g. rename some.ear some.ear.1 and rename it back to some.ear.1 to some .ear or cut/paste the ear to/from another directory then WebLogic can pick it up. It's not ideal but it works in case anyone else comes across the same issue.
I am newbie to Azure batch as well as docker. The problem i am facing is i have created an image based on another custom image in which some files and folders are created at the root level/directory of the container and every thing works fine but when the same image is running in Azure batch task, i dont know where these files and folders are being created because the wd (working directory) folder is empty. Any suggestions please? Thank you. I know the Azure batch does something with the directory structure but i am not clear about it.
As you're no doubt aware, Batch maps directories into the container (from the docs):
All directories recursively below the AZ_BATCH_NODE_ROOT_DIR (the root of Azure Batch directories on the node) are mapped into the container
so for example if you have a resource file on the task, this ends up in the working directory within the container. However this doesn't mean the container is allowed to write back to the same location on the host (but only within the container). I would suggest that you take whatever results/output you have generated and upload them into Azure Blob Storage via a Shared Access Signature - this is the usual way to get results from a Batch job even without using Docker.
I am trying to run Hyperledger's BYFN Tutorial on a Win10 Home using Docker Toolbox, with VirtualBox 5.2.4. I am using the default image for the VirtualBox VM.
I have set up a shared folder (not in C:/Users, but on my other drive) and it seems to be functioning correctly - changes I make from either Windows, or the docker-machine are reflected in both places as intended. I successfully generate the network artifacts using "./byfn -m generate", but I get an error when trying to "./byfn up" it.
What happens is that, as far as I can see from the logs, all the containers get brought up correctly, but for some reason the volumes of the cli container are not attached correctly (I think). When byfn.sh finishes I get the following error:
When I ssh into the cli container, I can see the channel-artifacts, crypto and scripts folders, but their contents don't seem to correlate with the volumes: part of the docker-compose file. First, the scripts folder is empty (whereas in the docker-compose file it's specified that it should mount a bunch of files), so I get the above error. Second, the channel-artifacts containes only 1 directory named genesis.block, which should actually be a file. And in the crypto folder there are just a bunch of directories.
As you might have guessed, I'm pretty new at docker, so this might be intended behavior, but I'm still getting an error.
Please let me know if I can provide additional information. Thanks in advance.