Reduce docker build context size

.dockerignore

@@ -1,325 +0,0 @@
-# huggingface cache
-.hf_cache/
-
-.idea/**/workspace.xml
-.idea/**/tasks.xml
-.idea/**/usage.statistics.xml
-.idea/**/dictionaries
-.idea/**/shelf
-.idea/**/contentModel.xml
-.idea/**/dataSources/
-.idea/**/dataSources.ids
-.idea/**/dataSources.local.xml
-.idea/**/sqlDataSources.xml
-.idea/**/dynamic.xml
-.idea/**/uiDesigner.xml
-.idea/**/dbnavigator.xml
-.idea/**/gradle.xml
-.idea/**/libraries
-cmake-build-*/
-.idea/**/mongoSettings.xml
-*.iws
-out/
-.idea_modules/
-atlassian-ide-plugin.xml
-.idea/replstate.xml
-com_crashlytics_export_strings.xml
-crashlytics.properties
-crashlytics-build.properties
-fabric.properties
-.idea/httpRequests
-.idea/caches/build_file_checksums.ser
-.Python
-[Bb]in
-[Ii]nclude
-[Ll]ib
-[Ll]ib64
-[Ll]ocal
-[Ss]cripts
-pyvenv.cfg
-.venv
-pip-selfcheck.json
-*.rsuser
-*.suo
-*.user
-*.userosscache
-*.sln.docstates
-*.userprefs
-mono_crash.*
-[Dd]ebug/
-[Dd]ebugPublic/
-[Rr]elease/
-[Rr]eleases/
-x64/
-x86/
-[Ww][Ii][Nn]32/
-[Aa][Rr][Mm]/
-[Aa][Rr][Mm]64/
-bld/
-[Bb]in/
-[Oo]bj/
-[Ll]og/
-[Ll]ogs/
-.vs/
-Generated\ Files/
-[Tt]est[Rr]esult*/
-[Bb]uild[Ll]og.*
-*.VisualState.xml
-TestResult.xml
-nunit-*.xml
-[Dd]ebugPS/
-[Rr]eleasePS/
-dlldata.c
-BenchmarkDotNet.Artifacts/
-project.lock.json
-project.fragment.lock.json
-artifacts/
-ScaffoldingReadMe.txt
-StyleCopReport.xml
-*_i.c
-*_p.c
-*_h.h
-*.ilk
-*.meta
-*.obj
-*.iobj
-*.pch
-*.pdb
-*.ipdb
-*.pgc
-*.pgd
-*.rsp
-*.sbr
-*.tlb
-*.tli
-*.tlh
-*.tmp
-*.tmp_proj
-*_wpftmp.csproj
-*.log
-*.vspscc
-*.vssscc
-.builds
-*.pidb
-*.svclog
-*.scc
-_Chutzpah*
-ipch/
-*.aps
-*.ncb
-*.opendb
-*.opensdf
-*.sdf
-*.cachefile
-*.VC.db
-*.VC.VC.opendb
-*.psess
-*.vsp
-*.vspx
-*.sap
-*.e2e
-$tf/
-*.gpState
-_ReSharper*/
-*.[Rr]e[Ss]harper
-*.DotSettings.user
-_TeamCity*
-*.dotCover
-.axoCover/*
-!.axoCover/settings.json
-coverage*.json
-coverage*.xml
-coverage*.info
-*.coverage
-*.coveragexml
-_NCrunch_*
-.*crunch*.local.xml
-nCrunchTemp_*
-*.mm.*
-AutoTest.Net/
-.sass-cache/
-[Ee]xpress/
-DocProject/buildhelp/
-DocProject/Help/*.HxT
-DocProject/Help/*.HxC
-DocProject/Help/*.hhc
-DocProject/Help/*.hhk
-DocProject/Help/*.hhp
-DocProject/Help/Html2
-DocProject/Help/html
-publish/
-*.[Pp]ublish.xml
-*.azurePubxml
-*.pubxml
-*.publishproj
-PublishScripts/
-*.nupkg
-*.snupkg
-**/[Pp]ackages/*
-!**/[Pp]ackages/build/
-*.nuget.props
-*.nuget.targets
-csx/
-*.build.csdef
-ecf/
-rcf/
-AppPackages/
-BundleArtifacts/
-Package.StoreAssociation.xml
-_pkginfo.txt
-*.appx
-*.appxbundle
-*.appxupload
-*.[Cc]ache
-!?*.[Cc]ache/
-ClientBin/
-~$*
-*~
-*.dbmdl
-*.dbproj.schemaview
-*.jfm
-*.pfx
-*.publishsettings
-orleans.codegen.cs
-Generated_Code/
-_UpgradeReport_Files/
-Backup*/
-UpgradeLog*.XML
-UpgradeLog*.htm
-ServiceFabricBackup/
-*.rptproj.bak
-*.mdf
-*.ldf
-*.ndf
-*.rdl.data
-*.bim.layout
-*.bim_*.settings
-*.rptproj.rsuser
-*- [Bb]ackup.rdl
-*- [Bb]ackup ([0-9]).rdl
-*- [Bb]ackup ([0-9][0-9]).rdl
-FakesAssemblies/
-*.GhostDoc.xml
-.ntvs_analysis.dat
-node_modules/
-*.plg
-*.opt
-*.vbw
-**/*.HTMLClient/GeneratedArtifacts
-**/*.DesktopClient/GeneratedArtifacts
-**/*.DesktopClient/ModelManifest.xml
-**/*.Server/GeneratedArtifacts
-**/*.Server/ModelManifest.xml
-_Pvt_Extensions
-.paket/paket.exe
-paket-files/
-.fake/
-.cr/personal
-__pycache__/
-*.pyc
-*.tss
-*.jmconfig
-*.btp.cs
-*.btm.cs
-*.odx.cs
-*.xsd.cs
-OpenCover/
-ASALocalRun/
-*.binlog
-*.nvuser
-.mfractor/
-.localhistory/
-healthchecksdb
-MigrationBackup/
-.ionide/
-FodyWeavers.xsd
-.DS_Store
-.AppleDouble
-.LSOverride
-Icon
-._*
-.DocumentRevisions-V100
-.fseventsd
-.Spotlight-V100
-.TemporaryItems
-.Trashes
-.VolumeIcon.icns
-.com.apple.timemachine.donotpresent
-.AppleDB
-.AppleDesktop
-Network Trash Folder
-Temporary Items
-.apdisk
-*.py[cod]
-*$py.class
-*.so
-build/
-develop-eggs/
-dist/
-downloads/
-eggs/
-.eggs/
-lib/
-lib64/
-parts/
-sdist/
-var/
-wheels/
-share/python-wheels/
-*.egg-info/
-.installed.cfg
-*.egg
-MANIFEST
-*.manifest
-*.spec
-pip-log.txt
-pip-delete-this-directory.txt
-htmlcov/
-.tox/
-.nox/
-.coverage
-.coverage.*
-.cache
-nosetests.xml
-coverage.xml
-*.cover
-*.py,cover
-.hypothesis/
-.pytest_cache/
-cover/
-*.mo
-*.pot
-local_settings.py
-db.sqlite3
-db.sqlite3-journal
-instance/
-.webassets-cache
-.scrapy
-docs/_build/
-.pybuilder/
-target/
-.ipynb_checkpoints
-profile_default/
-ipython_config.py
-__pypackages__/
-celerybeat-schedule
-celerybeat.pid
-*.sage.py
-.env
-env/
-venv/
-ENV/
-env.bak/
-venv.bak/
-.spyderproject
-.spyproject
-.ropeproject
-/site
-.mypy_cache/
-.dmypy.json
-dmypy.json
-.pyre/
-.pytype/
-cython_debug/
-*.md
-.git*

api.dockerignore

@@ -0,0 +1,8 @@
+.hf_cache/
+.idea
+*.md
+.git*
+.enditorconfig
+models
+converter
+tests

docker-compose.yaml

@@ -3,7 +3,7 @@ services:
   triton:
     build:
       context: .
-      dockerfile: Dockerfile
+      dockerfile: triton.Dockerfile
     command: bash -c "CUDA_VISIBLE_DEVICES=${GPUS} mpirun -n 1 --allow-run-as-root /opt/tritonserver/bin/tritonserver --model-repository=/model"
     shm_size: '2gb'
     volumes:
@@ -27,7 +27,7 @@ services:
     # For local build
     build:
       context: .
-      dockerfile: copilot_proxy/Dockerfile
+      dockerfile: proxy.Dockerfile
     env_file:
       # Automatically created via ./setup.sh
       - .env

tests/python_backend/docker-compose-with-gpus.yaml

@@ -3,7 +3,7 @@ services:
   triton:
     build:
       context: ../../
-      dockerfile: Dockerfile
+      dockerfile: ../../triton.Dockerfile
     command: bash -c "CUDA_VISIBLE_DEVICES="${GPUS}" mpirun -n 1 --allow-run-as-root /opt/tritonserver/bin/tritonserver --model-repository=/model"
     shm_size: '2gb'
     volumes:
@@ -27,7 +27,7 @@ services:
     # For local build
     build:
       context: ../../
-      dockerfile: copilot_proxy/Dockerfile
+      dockerfile: ../../proxy.Dockerfile
     env_file:
       # Automatically created via ./setup.sh
       - test.env

tests/python_backend/docker-compose-without-gpus.yaml

@@ -3,7 +3,7 @@ services:
   triton:
     build:
       context: ../../
-      dockerfile: Dockerfile
+      dockerfile: ../../triton.Dockerfile
     command: bash -c "CUDA_VISIBLE_DEVICES="${GPUS}" mpirun -n 1 --allow-run-as-root /opt/tritonserver/bin/tritonserver --model-repository=/model"
     shm_size: '2gb'
     volumes:
@@ -20,7 +20,7 @@ services:
     # For local build
     build:
       context: ../../
-      dockerfile: copilot_proxy/Dockerfile
+      dockerfile: ../../proxy.Dockerfile
     env_file:
       # Automatically created via ./setup.sh
       - test.env

tests/python_backend/requirements.txt

@@ -1,3 +1,3 @@
-pytest
+pytest==7.2.1
-pexpect
+pexpect==4.8.0
-requests
+requests==2.28.2

tests/python_backend/test_setup.py

@@ -1,9 +1,11 @@
-"Tests setup script (currently for Python backend)"
+"""
+Tests setup script (currently for Python backend)
+"""
 
 import os
-import subprocess
-import signal
 import shutil
+import signal
+import subprocess
 from pathlib import Path
 from typing import Dict, Union
 
@@ -14,30 +16,37 @@ import requests
 curdir = Path(__file__).parent
 root = curdir.parent.parent
 
-test_models_dir = curdir/"models"
+test_models_dir = curdir.joinpath("models")
 
 
 def setup_module():
-    "Setup steps for tests in this module"
+    """
-    assert (root/"setup.sh").exists(), "setup.sh not found"
+    Setup steps for tests in this module
-    if (root/".env").exists():
+    """
-        shutil.move(str(root/".env"), str(root/".env.bak"))
+    assert root.joinpath("setup.sh").exists(), "setup.sh not found"
+    if root.joinpath(".env").exists():
+        shutil.move(str(root.joinpath(".env")), str(root.joinpath(".env.bak")))
 
 def teardown_module():
-    "Teardown steps for tests in this module"
+    """
-    if (root/".env.bak").exists():
+    Teardown steps for tests in this module
-        shutil.move(str(root/".env.bak"), str(root/".env"))
+    """
+    if root.joinpath(".env.bak").exists():
+        shutil.move(str(root.joinpath(".env.bak")), str(root.joinpath(".env")))
     try:
-        if test_models_dir.exists():
+        if test_models_dir:
             shutil.rmtree(test_models_dir)
     except Exception as exc:
         print(
             f"WARNING: Couldn't delete `{test_models_dir}` most likely due to permission issues."
-            f"Run the tests with sudo to ensure this gets deleted automatically, or else delete manually. Exception: {exc}"
+            f"Run the tests with sudo to ensure this gets deleted automatically, or else delete manually. "
+            f"Exception: {exc}"
         )
 
 def enter_input(proc: pexpect.spawn, expect: str, input_s: str, timeout: int = 5) -> str:
-    "Helper function to enter input for a given prompt. Returns consumed output."
+    """
+    Helper function to enter input for a given prompt. Returns consumed output.
+    """
 
     try:
         proc.expect(expect, timeout=timeout)
@@ -53,7 +62,9 @@ def enter_input(proc: pexpect.spawn, expect: str, input_s: str, timeout: int = 5
     return after
 
 def run_common_setup_steps(n_gpus: int = 0) -> pexpect.spawn:
-    "Helper function to run common setup steps."
+    """
+    Helper function to run common setup steps.
+    """
     proc = pexpect.pty_spawn.spawn(
         "./setup.sh 2>&1", encoding="utf-8", cwd=str(root),
     )
@@ -68,10 +79,12 @@ def run_common_setup_steps(n_gpus: int = 0) -> pexpect.spawn:
     return proc
 
 def load_test_env():
-    "Load test env vars"
+    """
+    Load test env vars
+    """
     # Without loading default env vars, PATH won't be set correctly
     env = os.environ.copy()
-    with open(curdir/"test.env", "r", encoding="utf8") as test_env:
+    with open(curdir.joinpath("test.env"), "r", encoding="utf8") as test_env:
         for line in test_env:
             key, val = line.strip().split("=")
             env[key] = val
@@ -81,7 +94,9 @@ def run_inference(
     prompt: str, model: str = "py-model", port: int = 5000, return_all: bool = False,
     **kwargs
 ) -> Union[str, Dict]:
-    "Invokes the copilot proxy with the given prompt and returns the completion"
+    """
+    Invokes the copilot proxy with the given prompt and returns the completion
+    """
     endpoint = f"http://localhost:{port}/v1/engines/codegen/completions"
     data = {
         "model": model,
@@ -131,7 +146,7 @@ def test_python_backend(n_gpus: int):
     enter_input(proc, r".*run FauxPilot\? \[y/n\] ", "n", timeout=120)
 
     # copy $root/.env to $curdir/test.env
-    shutil.copy(str(root/".env"), str(curdir/"test.env"))
+    shutil.copy(str(root.joinpath(".env")), str(curdir.joinpath("test.env")))
 
     # run docker-compose up -f docker-compose-{without|with}-gpus.yml
     compose_file = f"docker-compose-with{'' if n_gpus > 0 else 'out'}-gpus.yaml"

triton.dockerignore

@@ -0,0 +1,8 @@
+.hf_cache/
+.idea
+*.md
+.git*
+.enditorconfig
+models
+converter
+tests