I'm following this tutorial and when I get to the part where I call:
cp /tf_files/stripped_retrained_graph.pb bazel-bin/tensorflow/examples/android/assets/stripped_output_graph.pb
and
cp /tf_files/retrained_labels.txt bin/tensorflow/examples/android/assets/imagenet_comp_graph_label_strings.txt
They both say "No such file or directory".
As you can see in this image I can cd to the tf_files folder and see that the files are there.
I can also cd to /tensorflow/tensorflow/examples/android/assets and call ls which shows there's just a BUILD file there.
In the cp command is there supposed to already be a stripped_output_graph.pb file in the destination which gets replaced? Or is it meant to just be creating a new file there?
Is there some way of doing cp [source] [current directory] rather than specifying the destination as a path?
I've tried removing the file path part in hope that it just uses the source filename but that doesn't work.
Calling
cp /tf_files/stripped_retrained_graph.pb /tensorflow/tensorflow/examples/android/assets/stripped_output_graph.pb
and
cp /tf_files/retrained_labels.txt /tensorflow/tensorflow/examples/android/assets/imagenet_comp_graph_label_strings.txt
finally worked, wasn’t at all obvious that I’d have to change the destination path or what it should be though.
Also I accidentally saved a file as .p rather than .pb but managed to remove it using $ docker exec <container> rm -rf /tensorflow/tensorflow/examples/android/asset
s/stripped_output_graph.p
Now I managed to copy the files in correctly, but then when I installed the app it was still just running the regular demo app.
Not sure why it didn’t work, so frustrating.
When I rebuilt it after copying the files in I got these conflict messages
Are these normal to have?
It looks like maybe a different labels file is taking priority over mine, how can I reach the external/inception5h/imagenet_comp_graph_label_strings.txt file to delete it so my file is used instead?
Does the “external” part mean that I can’t actually access it?
Related
I have a container that I want to export as a .tar file. I have used a podman run with a tar --exclude=/dir1 --exclude=/dir2 … that outputs to a file located on a bind-mounted host dir. But recently this has been giving me some tar: .: file changed as we read it errors, which podman/docker export would avoid. Besides the export I suppose is more efficient. So I'm trying to migrate to using the export, but the major obstacle is I can't seem to find a way to exclude paths from the tar stream.
If possible, I'd like to avoid modifying a tar archive already saved on disk, and instead modify the stream before it gets saved to a file.
I've been banging my head for multiple hours, trying useless advices from ChatGPT, looking at cpio, and attempting to pipe the podman export to tar --exclude … command. With the last I did have small success at some point, but couldn't make tar save the result to a particularly named file.
Any suggestions?
(note: I do not make distinction between docker and podman here as their export command is completely the same, and it's useful for searchability)
I am trying to backup jenkins home directory (/home/ubuntu/.jenkins/) using rsync to the target directory /opt/jenbkup/. Since the directory traversal seems not working as expected, I have gone with single directory in the filter:
rsync -avr --include="jobs/*/config.xml" --exclude="*" /home/saga/.jenkins /opt/jenbkup
But nothing is copied. I also tried with exact file path in the include and did not work.
rsync -avr --include="jobs/job1/config.xml" --exclude="*" /home/saga/.jenkins /opt/jenbkup
File is not copied to destination. I don't understand whats wrong here. Some one please assist.
I assume, that you would only backup the config.xml files in your JENKINS_HOME, then this should work:
rsync -av --include="*/" --include="config.xml" --exclude="*" \
--delete --prune-empty-dirs /home/saga/.jenkins/ /opt/jenbkup/
Short explaination of the used options:
--include="*/" traverse all directories
--include="config.xml" include only files named "config.xml"
--exclude="*" exlude everything
--delete delete non-existing files in the backup
--prune-empty-dirs delete empty directories from the backup
So I am trying to make a simple docker container to run sabnzbd in a container.
I do a git clone and then move sabnzbd folder into container.
Note: When I look in sabnzbd, all files look ok so it's not a git problem or branch etc. I'm happy so far there.
When I then run the container, I double checked but half the files are missing. For example, I was looking to make sure cheerypy was copied but it wasn't, and yes, i can confirm that I double checked again in sabnzbd folder.
So I have a folder with Dockerfile and sabnzbd. I built from that folder with the command:
sudo docker build --no-cache=true -t sabnzbd -f Dockerfile .
In a nutshell, the only thing that worked was Copy . /
I tried COPY sabnzbd/* sabnzbd and other variations.
I originally thought it was getting files from elsewhere but removed any trace of sabnzbd, but for my understanding it only looks at files relative to the docker build file.
I just thought this was so soo weird and wanted to get some thoughts even though I fixed it. I did remove all images and started from scratch but result was the same.
I did try the no cache option with build but still the same.
Thoughts?
A docker add will nicely extract the supplied compressed file into the directory specified in the zip/tar file
How can I extract it into a different directory?
Eg. if the file extracts to /myfile but I would prefer /otherFile
Don't believe there's any way to do this just using the ADD instruction. ADD supports a target directory obviously, like ADD ["<src>", "<dest>"] however it's still going to extract into the dir you have in the tar within that.
2 options, either rename the dir in the tar or do a RUN mv myfile otherfile after adding.
Is there a specific reason you need it to be named something in particular?
Think about this scenario where you build a tomcat image,
ADD apache-tomcat-8.0.48.tar.gz /opt
This cmd will extract the tar to /opt/apache-tomcat-8.0.48 , if you don't like the long folder name(apache-tomcat-8.0.48) then the requirement happens.
I'm having an issue with docker-compose where I'm passing a file into the container when it's run. The issue is that it doesn't seem to recognize when the file has been changed and serves the saved result back indefinitely until I change the name of the file.
An example (modified names for brevity):
jono#macbook:~/myProj% docker-compose run vpn conf.opvn
Options error: Unrecognized option or missing parameter(s) in conf.opvn:71: AXswRE+
5aN64mYiPSatOACC6+bISv8RcDPX/lMYdLwe8zQY6qWtbrjFXrp2 (2.3.8)
Then I change the file, save it, and run the command again - exact same output.
Then without changing anything I do this:
jono#macbook:~/myProj% cp conf.opvn newconf.opvn
And when I run $ docker-compose run vpn newconf.opvn it works. Seems really silly.
I'm working with Tmux and Mac if there is some way that affects it. Is this the expected behaviour? I couldn't find anything documenting this on the docker-compose homepage.
EDIT:
Specifically I'm using this repo from the amazing Jess.
The image you are using is using volume in order to mount your current directory. Basically the file conf.opvn is copied to the docker container.
When you change the file, the container doesn't see that change, but it does pick up the rename (which the container sees as a new file). This most probably is due to user rights of the file and the user rights of the folder in the docker container where this file is mounted. Try changing the file's permissions to 777 before beginning the process and check again.
You can find a discussion about this in the official forum of docker