Reputation
Badges 1
533 × Eureka!How do I get from the node to the task object?
Okay, so let me get this straight
The autoscaling is basically an ever-running task (lets say on the services
queue). Now, the actual auto scaling and which queues exist have nothign to do with that, and are configured in the auto scale task?
Oh... from the docs I understood that I don't have to run the script, that I can either configure it in the UI, or with the sscript (wizard) so I ignored it up until now
I was here, but I can't find info for the questions I mentioned
We try to break up every thing into independent tasks and group them using a pipeline. The dependency on an agnet caused an unnecessary overhead since we just want to execute locally. It became a burden once new data scientists join the project and instead of just telling them "yeah, just execute this script" you have to now teach them about clearml, the role of agents, how to launch them, how they behave, how to remove them and stuff like that... things you want to avoid with data scientists
let me try to docker-compose down --rmi all
not manually I assume that if I deleted the image, and then docker-composed up, and I can see the pull working it should pull the correct one
I have them in two different places, once under Hyperparameters -> General
I was trying out the pipeline controller for the first time and I felt a bit of a burden that just for the sake of trying I had to launch an agent
can't remember, I just restarted everything so I don't have this info now
The latest, I curl
ed the docker-compose like 10 minutes ago
I set it to true and restarted by agent
btw my site packages is false - should it be true? You pasted that but I'm not sure what it should be, in the paste is false but you are asking about true
it seems that only the packages that are on the script are getting installed
` alabaster==0.7.12
appdirs==1.4.4
apturl==0.5.2
attrs==21.2.0
Babel==2.9.1
bcrypt==3.1.7
blinker==1.4
Brlapi==0.7.0
cachetools==4.0.0
certifi==2019.11.28
chardet==3.0.4
chrome-gnome-shell==0.0.0
clearml==1.0.5
click==8.0.1
cloud-sptheme==1.10.1.post20200504175005
cloudpickle==1.6.0
colorama==0.4.3
command-not-found==0.3
cryptography==2.8
cupshelpers==1.0
cycler==0.10.0
Cython==0.29.24
dbus-python==1.2.16
decorator==4.4.2
defer==1.0.6
distlib==0.3.1
distro==1.4.0
distro-info===0.23ubuntu1
doc...
This is the pip freeze
of the environment I don't know why it differs from what the agent has... the agent only has a subset of these google libs
the output above is what the agent has as it seems... obviously on my machine I have it installed
TimelyPenguin76
In the larger context I'd look on how other object stores treat similar problems, I'm not that advanced in these topics.
But adding a simple force_download
flag to the get_local_copy
method could solve many cases I can think of, for example I'd set it to true in my case as I don't mind the times it will re-download when not necessary as it is quite small (currently I always delete the local file, but it looks pretty ugly)
Continuing on this line of thought... Is it possible to call task.execute_remotely
on a CPU only machine (data scientists' laptop for example) and make the agent that fetches this task to run it using GPU? I'm asking that because it is mentioned that it replicates the running environment on the task creator... which is exactly what I'm not trying to do 😄