compare between both of the tasks

same basic job not gets overwritten, but created new one every time

AgitatedDove14 indeed there are few sub projects
do you suggest to delete those first?

I think there are some experiments that are messing up mongodb
this logs unusual in clearml-mongo logs:

{"t":{"$date":"2023-09-19T12:15:50.685+00:00"},"s":"I",  "c":"COMMAND",  "id":51803,   "ctx":"conn73","msg":"Slow query","attr":{"type":"command","ns":"backend.model","command":{"distinct":"model","key":"project","query":{"$and":[{"$or":[{"company":{"$in":["d1bd92a3b039400cbafc60a7a5b1e52b",null,""]}},{"company":{"$exists":false}}]},{"user":{"$in":["197aea8467d3f471fc0db98b57ed80fa"]...

unfortunately I couldn't fix this
the ES state in hectic, can't delete anything
clearml is still live, read-only mode, all existing indices are readable
new jobs can't write to this clearml server

I manage clearml-server for many users
wrote a script that pulls tasks (for review - pending deletion) and limit for specific user (saw the REST can limit only by user id )
so user will need to provide the script his own user id in advance
I guess my options are:

• maintain stastic list of users (which I get from users.get_all endpoint)
• add to my script another query that will match user name to user id, before running the tasks_get.all for specific user id

yep, again most jobs works .. the issue with when a job tries to upload artifacts to fileserver

yep, that's options number 2 🙂
I will implement that

trying to use projects.get_all to pull all my projects into single file
and there are more then 500 ...

hmmm that's not it, same endless loop when using body

I know the 500 limit and using it

but my while loop keeps pulling the same 500 ... and running endless

OK I got everything to work
I think this script can be useful to other people and will be happy to share

@<1523701070390366208:profile|CostlyOstrich36> is there some repo I fork and contribute?

VivaciousPenguin66 your docs was helpful, I got SSL running but my question remained
have you kept needed http services accessible and only running the authentication via https?
api_server: "http://<my-clearml-server>:8008" web_server: " " files_server: "http://<my-clearml-server:8081"my current state is that the webserver is accessible via http and https, in 8080 & 443

ohhh severe error here 🙂
I was mixed between other API I worked on .. and did not read carefully the right flag

simple adding page to the body did the work
thanks again @<1724235687256920064:profile|LonelyFly9>

you are correct and thank you for the reply @<1523701070390366208:profile|CostlyOstrich36>
going forward, I assume the clearml-server open-source releases will be continue to be released in Docker Hub

from clearml import Task task = Task.init(project_name="Inbar2022/LanguageFactoryDanish/lions_test", task_name="lions3")
python main.py --cuda --epoch 1

@<1523701087100473344:profile|SuccessfulKoala55> seen this with 1.13.2
worth to try and upgrade to latest?

I'd guess mongo is choking, not sure why

when running in debug and watch the values I get

    data = response.json()
    projects = data['data']['projects']
    all_data.extend(projects)

in each loop iteration
projects are same 500 values
all_data gets append for same 500 values in endless loop

I have bug in my code and can't find where just yet

I think this is the right approach, let me have a deeper look
thanks @<1724235687256920064:profile|LonelyFly9>

ok, hopefully someone will share some thoughts and how it went 🙂

thanks @<1523701070390366208:profile|CostlyOstrich36>
I've done this successfully using the API already
as for the sdk option - in which format should I provide the list of tasks/projects to the sdk?

for
only_fields=["id", "name","created","status_changed","status", "user"], :

output example

{'id': '02a3f5929cf246138994c9243a692219', 'name': 'docfm_v7_safe_32gpu80g_11Jan24_4w', 'created': datetime.datetime(2024, 1, 11, 9, 54, 33, 406000, tzinfo=tzutc()), 'status_changed': dateti...

looking into ES index events-training_stats_scalar-d1bd92a3b039400cbafc60a7a5b1e52b

docs.count docs.deleted store.size pri.store.size
2118131043     29352476    265.1gb        265.1gb

sounds we're hitting some ES limitation?

not really - I can try to run these in parallel

so I think I'm in the right direction
adding verify= and pointing to my CA.pem looks like the right approach

now, how do I use it with ClearML API?

cleanup_service

for task in tasks:
    try:
      deleted_task = Task.get_task(task_id=task.id)
      print (deleted_task.name)
      deleted_task.delete(
               delete_artifacts_and_models=True,
               skip_models_used_by_other_tasks=True,
               raise_on_error=False
      )

it throw down the SSL error,...

@<1523701070390366208:profile|CostlyOstrich36> unfortunately, this is not the behavior we are seeing
same exact issue happen tonight
on epoch number 53 ClearML were shut down, the job did not continue to epoch 54 and eventually got killed with watchdog timer

correct

in the UI I also see the display name, so I pulled all the users info, and match name to id

how do you know the ID?
I'm looking for this ...

so I have large json, with list of task id's
which I want to delete in bulk

API is doable

how about the sdk? how do I provide a list of tasks id's to for deletion

from the cleanup example:

        for task in tasks:
            try:
                deleted_task = Task.get_task(task_id=task.id)
                deleted_task.delete(

how do I set tasks , while coming from known list of task id's