ClearML/clearml-agent
1
0
Fork 0
mirror of https://github.com/clearml/clearml-agent synced 2025-02-26 05:59:24 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity
800 Commits 2 Branches 80 Tags 9.5 MiB
66494c598d
Commit Graph

435 Commits

Author SHA1 Message Date
clearml
66494c598d Fix on_abort bash callback, if main processes leave while on_abort callback is running, wait for the on_abort to complete 2025-02-24 13:46:55 +02:00
clearml
d30d4e7e61 Fix session should retry on Any error if send fails 2025-02-24 13:46:29 +02:00
clearml
0e2657421f Added CLEARML_AGENT_ABORT_CALLBACK_CMD and CLEARML_AGENT_ABORT_CALLBACK_TIMEOUT 
(default 180 sec) to define callback command to be called on abort status change
 2025-02-24 13:46:00 +02:00
clearml
ee286e2fb7 Fix container default arguments should never be a list 2025-02-24 13:44:52 +02:00
clearml
d87521c36c Add support for container rulebook overrides ('force_container_rules: true') and container rulebook task update ('update_back_task: true'). 
This addition allows users to override container arguments forcefully based on the tasks properties (repo, tags, project, user etc.), as well as offer additional defaults based on python required packages or python versions
 2025-02-24 13:44:26 +02:00
clearml
8887453328 Cleanup error prints on bash startup script 2025-02-24 13:42:37 +02:00
clearml
8d3cb34390 Add default support for dns i.e. rocky/centos/fedora containers 2025-02-24 13:41:32 +02:00
clearml
528bf314ef Update GpuFractionsHandler GPU name to mem size 2025-02-24 13:30:30 +02:00
clearml
4f91c45d38 Fix untitled file based on binary is now py/sh based on requested binary 2025-02-24 13:29:56 +02:00
clearml
0a13fd79fc Make sure that if we fail to kill a child processes we continue to try the rest 2025-02-24 13:26:49 +02:00
clearml
c9f5b3d19a Force the stop command to avoid a potential race 2025-02-24 13:26:26 +02:00
clearml
4219835aa1 Fix pip requirements print dump should be sorted 2025-02-24 13:25:23 +02:00
clearml
d32b82cb01 Integrate docker port mapping, to control non network=host port mapping, including port reassigning for multiple running agents on the same machine 2025-02-24 13:24:58 +02:00
clearml
97cb47d48e Add docker port mapping parsing and reassigning feature support 
Add initial component import from clearml-sdk for easier integration
 2025-02-24 13:24:35 +02:00
clearml
8f28d2882a Fix pip freeze dump to comply with yaml fancy print 2025-02-24 13:23:50 +02:00
clearml
546ffff95d Fix cached venv tried to reinstall priority packages even through they are preinstalled 2025-02-24 13:23:00 +02:00
clearml
a6ae6b2095 Add initial support for --break-system-packages version detection, but the reality is that we do not need it, because by the time we are running it is too late, so we do rm /usr/lib/python3.*/EXTERNALLY-MANAGED 2025-01-26 23:07:09 +02:00
clearml
369b440b96 Reduce required packages 2025-01-26 23:05:57 +02:00
clearml
28e9280a4f Reduce required packages 2025-01-26 23:03:16 +02:00
clearml
7e9e3ad08b Add printout when using custom configuration file 2025-01-26 22:51:09 +02:00
clearml
44709673f4 Add CLEARML_AGENT_CONFIG_VERBOSE for verbose configuration file loading 2025-01-26 22:50:49 +02:00
clearml
4158146420 Version bump to v1.9.3 2025-01-19 16:17:56 +02:00
clearml
91dfa09466 Fix Python 3.13 support 2025-01-05 12:14:24 +02:00
clearml
070919973b Fix python 3.6 compatibility, no := operator 2025-01-05 12:13:21 +02:00
clearml
47d35ef48f Fix managed python environment inside container (PEP 668) remove usr/lib/python3.*/EXTERNALLY-MANAGED 2024-12-26 18:59:42 +02:00
clearml
54ed234fca Add agent.docker_args_filters to configuration docs 2024-12-26 18:58:58 +02:00
clearml
a26860e79f Fix default value handling in merge_dicts() 2024-12-26 18:58:24 +02:00
clearml
fc1abbab0b Refactor k8s glue 2024-12-26 18:58:00 +02:00
clearml
4fa61dde1f Support ignoring kubectl errors 2024-12-12 23:41:31 +02:00
clearml
26d748a4d8 Support creating queue with tags 2024-12-12 23:40:57 +02:00
clearml
d8366dedc6 Fix UV priority 
Fix UV cache is disabled, UV handles its own cache
Fix UV freeze
Fix make sure we do not use pip cache if poetry/uv is used (even if we reverted to pip we can't know if someone changed the repository and now in a new version, a lock file exists)
 2024-12-12 23:38:42 +02:00
mads-oestergaard
cc656e2969
Add support for uv as package manager (#218) 
* add uv as a package manager

* update configs

* update worker and defs

* update environ

* Update configs to highlight sync command

* rename to sync_extra_args and set UV_CACHE_DIR
 2024-11-27 13:44:55 +02:00
clearml
b65e5fed94 Scan more Python 3 versions 2024-11-17 13:55:51 +02:00
clearml
3273f76b46 Version bump to v1.9.2 2024-10-28 18:33:04 +02:00
clearml
9af0f9fe41 Fix reload method is found in the config object 2024-10-28 18:12:22 +02:00
clearml
205cd47cb9 Fix use req_token_expiration_sec when creating a task session and not the default value 2024-10-28 18:11:42 +02:00
clearml
0ff428bb96 Fix report index not advancing in resource monitoring causes more than one GPU not to be reported 2024-10-28 18:11:00 +02:00
Matteo Destro
bf8d9c96e9
Handle OSError when checking for is_file (#215) 2024-10-13 10:08:03 +03:00
allegroai
a88487ff25 Add support for pip legacy resolver for versions specified in the agent.package_manager.pip_legacy_resolver configuration option 
Add skip existing packages
 2024-09-22 22:36:06 +03:00
Jake Henning
785e22dc87
Version bump to v1.9.1 2024-09-02 01:04:49 +03:00
Jake Henning
6a2b778d53
Add default pip version support for Python 3.12 2024-09-02 01:03:52 +03:00
allegroai
b2c3702830 Version bump to v1.9.0 2024-08-28 23:18:26 +03:00
allegroai
6302d43990 Add support for skipping container apt installs using CLEARML_AGENT_SKIP_CONTAINER_APT env var in k8s 
Add runtime callback support for setting runtime properties per task in k8s
Fix remove task from pending queue and set to failed when kubectl apply fails
 2024-08-27 23:01:27 +03:00
allegroai
760bbca74e Fix failed Task in services mode logged "User aborted" instead of failed, add Task reason string 2024-08-27 22:56:37 +03:00
allegroai
e63fd31420 Fix string format 2024-08-27 22:55:49 +03:00
allegroai
2ff9985db7 Add user ID to the vault loading print 2024-08-27 22:55:32 +03:00
allegroai
b8c762401b Fix use same state transition if supported by the server (instead of stopping the task before re-enqueue) 2024-08-27 22:54:45 +03:00
allegroai
99e1e54f94 Add support for tasks containing only bash script or python module command 2024-08-27 22:53:14 +03:00
allegroai
a4d3b5bad6 Fix only set Task started status on node rank 0 2024-08-27 22:52:31 +03:00
allegroai
b21665ed6e Fix do not cache venv cache if venv/python skip env var was set 2024-08-27 22:52:01 +03:00