D6922: ci: implement a "try server"

indygreg (Gregory Szorc) phabricator at mercurial-scm.org
Mon Sep 30 00:54:16 UTC 2019


indygreg created this revision.
Herald added a subscriber: mercurial-devel.
Herald added a reviewer: hg-reviewers.

REVISION SUMMARY
  This commit implements a new CI component which allows privileged users
  to upload Mercurial bundles and have automation run against them. The
  functionality is similar to what exists in `contrib/automation/automation.py`
  for running tests on remote AWS infrastructure. The big difference is
  most of the heavy work is handled by the CI system instead of by the
  client invoking AWS APIs directly. This makes it easier to scale and
  doesn't require the user to have substantial privileges for an AWS
  account.

REPOSITORY
  rHG Mercurial

REVISION DETAIL
  https://phab.mercurial-scm.org/D6922

AFFECTED FILES
  contrib/ci/README.rst
  contrib/ci/ci.py
  contrib/ci/hgci/__init__.py
  contrib/ci/hgci/cli.py
  contrib/ci/hgci/try_server.py
  contrib/ci/lambda_functions/ci.py
  contrib/ci/lambda_functions/web.py
  contrib/ci/requirements.txt
  contrib/ci/requirements.txt.in
  contrib/ci/terraform/iam_users.tf
  contrib/ci/terraform/repo_change_reactor.tf
  contrib/ci/terraform/storage.tf
  contrib/ci/terraform/try_server.tf
  contrib/ci/terraform/worker.tf
  tests/test-check-code.t

CHANGE DETAILS

diff --git a/tests/test-check-code.t b/tests/test-check-code.t
--- a/tests/test-check-code.t
+++ b/tests/test-check-code.t
@@ -20,6 +20,8 @@
   Skipping contrib/automation/hgautomation/ssh.py it has no-che?k-code (glob)
   Skipping contrib/automation/hgautomation/windows.py it has no-che?k-code (glob)
   Skipping contrib/automation/hgautomation/winrm.py it has no-che?k-code (glob)
+  Skipping contrib/ci/hgci/cli.py it has no-che?k-code (glob)
+  Skipping contrib/ci/hgci/try_server.py it has no-che?k-code (glob)
   Skipping contrib/ci/lambda_functions/ci.py it has no-che?k-code (glob)
   Skipping contrib/ci/lambda_functions/web.py it has no-che?k-code (glob)
   Skipping contrib/packaging/hgpackaging/downloads.py it has no-che?k-code (glob)
diff --git a/contrib/ci/terraform/worker.tf b/contrib/ci/terraform/worker.tf
--- a/contrib/ci/terraform/worker.tf
+++ b/contrib/ci/terraform/worker.tf
@@ -13,6 +13,12 @@
 }
 
 data "aws_iam_policy_document" "ci_worker" {
+  # Allow CI worker EC2 instances to fetch try bundles from S3.
+  statement {
+    effect = "Allow"
+    actions = ["s3:GetObject"]
+    resources = ["${aws_s3_bucket.mercurial_try_bundles.arn}/*"]
+  }
   # Allow CI worker EC2 instances to write artifacts to S3.
   statement {
     effect = "Allow"
diff --git a/contrib/ci/terraform/try_server.tf b/contrib/ci/terraform/try_server.tf
new file mode 100644
--- /dev/null
+++ b/contrib/ci/terraform/try_server.tf
@@ -0,0 +1,108 @@
+# Holds user-uploaded Mercurial bundles to trigger CI on.
+resource "aws_s3_bucket" "mercurial_try_bundles" {
+  bucket = "mercurial-try-bundles"
+  region = "us-west-2"
+  acl = "private"
+
+  lifecycle_rule {
+    id = "Purge old objects"
+    enabled = true
+    expiration {
+      days = 30
+    }
+  }
+}
+
+# Logging for Lambda function.
+resource "aws_cloudwatch_log_group" "lambda_ci_try_server_upload" {
+  name = "/aws/lambda/${aws_lambda_function.ci_try_server_upload.function_name}"
+  # Longer than other log groups for auditing purposes.
+  retention_in_days = 30
+}
+
+# Role for Lambda function.
+resource "aws_iam_role" "lambda_ci_try_server_upload" {
+  name = "lambda-ci-try-server-upload"
+  description = "For Lambaa function handling Try server uploads"
+  assume_role_policy = data.aws_iam_policy_document.assume_role_lambda.json
+}
+
+# The Lambda function which handles bundle upload and triggering jobs.
+resource "aws_lambda_function" "ci_try_server_upload" {
+  function_name = "ci-try-server-upload"
+  description = "Handles uploading of bundles to the Try server"
+  filename = data.archive_file.lambda_ci.output_path
+  handler = "ci.handle_try_server_upload"
+  source_code_hash = data.archive_file.lambda_ci.output_base64sha256
+  runtime = "python3.7"
+  timeout = 60
+  role = aws_iam_role.lambda_ci_try_server_upload.arn
+  environment {
+    variables = {
+      DYNAMODB_PUSH_TABLE = aws_dynamodb_table.ci_push.name
+      S3_TRY_BUNDLES_BUCKET = aws_s3_bucket.mercurial_try_bundles.bucket
+      WEB_URL = "https://${aws_api_gateway_domain_name.ci_web.domain_name}/"
+    }
+  }
+}
+
+data "aws_iam_policy_document" "ci_try_server_upload" {
+  # Allow Lambda function to write CloudWatch events.
+  statement {
+    effect = "Allow"
+    actions = [
+      "logs:CreateLogGroup",
+      "logs:CreateLogStream",
+      "logs:PutLogEvents",
+    ]
+    resources = [aws_cloudwatch_log_group.lambda_ci_try_server_upload.arn]
+  }
+  # Enable writing bundle to S3 bucket.
+  statement {
+    effect = "Allow"
+    actions = [
+      "s3:PutObject",
+      "s3:PutObjectAcl",
+    ]
+    resources = ["${aws_s3_bucket.mercurial_try_bundles.arn}/*"]
+  }
+  # Enable querying and inserting pushes into DynamoDB.
+  statement {
+    effect = "Allow"
+    actions = [
+      "dynamodb:PutItem",
+      "dynamodb:Query",
+    ]
+    resources = [
+      aws_dynamodb_table.ci_push.arn,
+      "${aws_dynamodb_table.ci_push.arn}/*",
+    ]
+  }
+}
+
+resource "aws_iam_role_policy" "lambda_ci_try_server_upload" {
+  role = aws_iam_role.lambda_ci_try_server_upload.name
+  name = aws_iam_role.lambda_ci_try_server_upload.name
+  policy = data.aws_iam_policy_document.ci_try_server_upload.json
+}
+
+# An IAM group where membership will grant permission to invoke
+# Lambda function.
+resource "aws_iam_group" "access_to_try_upload" {
+  name = "access-to-try-upload"
+}
+
+data "aws_iam_policy_document" "access_to_try_upload" {
+  # Allow invoking the try upload Lambda function.
+  statement {
+    effect = "Allow"
+    actions = ["lambda:InvokeFunction"]
+    resources = [aws_lambda_function.ci_try_server_upload.arn]
+  }
+}
+
+resource "aws_iam_group_policy" "access_to_try_upload" {
+  name = aws_iam_group.access_to_try_upload.name
+  group = aws_iam_group.access_to_try_upload.name
+  policy = data.aws_iam_policy_document.access_to_try_upload.json
+}
diff --git a/contrib/ci/terraform/storage.tf b/contrib/ci/terraform/storage.tf
--- a/contrib/ci/terraform/storage.tf
+++ b/contrib/ci/terraform/storage.tf
@@ -25,10 +25,23 @@
     name = "push_id"
     type = "S"
   }
+  attribute {
+    name = "node"
+    type = "S"
+  }
 
   hash_key = "repo"
   range_key = "push_id"
 
+  # This allows us to easily query for whether a push already exists for
+  # a specific node in a repository.
+  global_secondary_index {
+    name = "repo-node"
+    hash_key = "repo"
+    range_key = "node"
+    projection_type = "KEYS_ONLY"
+  }
+
   stream_enabled = true
   stream_view_type = "NEW_AND_OLD_IMAGES"
 }
diff --git a/contrib/ci/terraform/repo_change_reactor.tf b/contrib/ci/terraform/repo_change_reactor.tf
--- a/contrib/ci/terraform/repo_change_reactor.tf
+++ b/contrib/ci/terraform/repo_change_reactor.tf
@@ -71,7 +71,6 @@
     ]
     resources = ["*"]
   }
-  # Allow querying and recording job state in DynamoDB.
   statement {
     effect = "Allow"
     actions = [
diff --git a/contrib/ci/terraform/iam_users.tf b/contrib/ci/terraform/iam_users.tf
new file mode 100644
--- /dev/null
+++ b/contrib/ci/terraform/iam_users.tf
@@ -0,0 +1,13 @@
+# This file defines AWS account users.
+
+resource "aws_iam_user" "gps" {
+  name = "gps"
+}
+
+resource "aws_iam_group_membership" "access_to_try_upload" {
+  name = "Membership to the access-to-try group to allow Try Server access"
+  group = aws_iam_group.access_to_try_upload.name
+  users = [
+    aws_iam_user.gps.name,
+  ]
+}
diff --git a/contrib/ci/requirements.txt.in b/contrib/ci/requirements.txt.in
new file mode 100644
--- /dev/null
+++ b/contrib/ci/requirements.txt.in
@@ -0,0 +1 @@
+boto3
diff --git a/contrib/ci/requirements.txt b/contrib/ci/requirements.txt
new file mode 100644
--- /dev/null
+++ b/contrib/ci/requirements.txt
@@ -0,0 +1,38 @@
+#
+# This file is autogenerated by pip-compile
+# To update, run:
+#
+#    pip-compile --generate-hashes --output-file=contrib/ci/requirements.txt contrib/ci/requirements.txt.in
+#
+boto3==1.9.238 \
+    --hash=sha256:2fc1c407a5ab08cfcf54eb4171d85c523bd27019ab890de257d018af2770f71d \
+    --hash=sha256:c215cf2c8e5e7b28ae7544b1cbdbc3216bef983d7adb8b701a64f9b893e0320b
+botocore==1.12.238 \
+    --hash=sha256:1ca993f0dc70591e0fca6cf3837ee9be52fd4fbbf1aa96ba1d4a860b41f676b7 \
+    --hash=sha256:6ec3297b87d3e2c4d88b009f91061aaecdb2ceef6d9be9386394571353909adb \
+    # via boto3, s3transfer
+docutils==0.15.2 \
+    --hash=sha256:6c4f696463b79f1fb8ba0c594b63840ebd41f059e92b31957c46b74a4599b6d0 \
+    --hash=sha256:9e4d7ecfc600058e07ba661411a2b7de2fd0fafa17d1a7f7361cd47b1175c827 \
+    --hash=sha256:a2aeea129088da402665e92e0b25b04b073c04b2dce4ab65caaa38b7ce2e1a99 \
+    # via botocore
+jmespath==0.9.4 \
+    --hash=sha256:3720a4b1bd659dd2eecad0666459b9788813e032b83e7ba58578e48254e0a0e6 \
+    --hash=sha256:bde2aef6f44302dfb30320115b17d030798de8c4110e28d5cf6cf91a7a31074c \
+    # via boto3, botocore
+python-dateutil==2.8.0 \
+    --hash=sha256:7e6584c74aeed623791615e26efd690f29817a27c73085b78e4bad02493df2fb \
+    --hash=sha256:c89805f6f4d64db21ed966fda138f8a5ed7a4fdbc1a8ee329ce1b74e3c74da9e \
+    # via botocore
+s3transfer==0.2.1 \
+    --hash=sha256:6efc926738a3cd576c2a79725fed9afde92378aa5c6a957e3af010cb019fac9d \
+    --hash=sha256:b780f2411b824cb541dbcd2c713d0cb61c7d1bcadae204cdddda2b35cef493ba \
+    # via boto3
+six==1.12.0 \
+    --hash=sha256:3350809f0555b11f552448330d0b52d5f24c91a322ea4a15ef22629740f3761c \
+    --hash=sha256:d16a0141ec1a18405cd4ce8b4613101da75da0e9a7aec5bdd4fa804d0e0eba73 \
+    # via python-dateutil
+urllib3==1.25.6 \
+    --hash=sha256:3de946ffbed6e6746608990594d08faac602528ac7015ac28d33cee6a45b7398 \
+    --hash=sha256:9a107b99a5393caf59c7aa3c1249c16e6879447533d0887f4336dde834c7be86 \
+    # via botocore
diff --git a/contrib/ci/lambda_functions/web.py b/contrib/ci/lambda_functions/web.py
--- a/contrib/ci/lambda_functions/web.py
+++ b/contrib/ci/lambda_functions/web.py
@@ -66,11 +66,21 @@
         '</style>',
     ]
 
-    for repo_entry in repo_poll_table.scan(Select='ALL_ATTRIBUTES')['Items']:
+    repos = repo_poll_table.scan(Select='ALL_ATTRIBUTES')['Items']
+
+    # Try repository is virtual. Add it manually.
+    repos.append({
+        'repo': 'try',
+    })
+
+    for repo_entry in repos:
         repo_name = repo_entry['repo']
-        repo_url = repo_entry['repo_url']
+        repo_url = repo_entry.get('repo_url', None)
 
-        html.append('<h1><a href="%s">%s</a></h1>' % (e(repo_url, quote=True), e(repo_name)))
+        if repo_url:
+            html.append('<h1><a href="%s">%s</a></h1>' % (e(repo_url, quote=True), e(repo_name)))
+        else:
+            html.append('<h1>%s</h1>' % e(repo_name))
 
         res = push_table.query(
             KeyConditionExpression=Key('repo').eq(repo_name),
@@ -401,9 +411,14 @@
 def push_info(push, repo_url):
     cset_url = '%s/rev/%s' % (repo_url, push['node'])
 
+    if repo_url:
+        cset_entry = '<a href="%s">%s</a>' % (
+            e(cset_url, quote=True), e(push['node']))
+    else:
+        cset_entry = e(push['node'])
+
     return ''.join([
-        '<h2>Changeset <span class="mono"><a href="%s">%s</a></span></h2>' % (
-            e(cset_url, quote=True), e(push['node'])),
+        '<h2>Changeset <span class="mono">%s</span></h2>' % cset_entry,
         '<p>branch: <span class="mono">%s</span></p>' % e(push['branch']),
         '<p>author: <span class="mono">%s</span></p>' % e(push['user']),
         '<p>description: <span class="mono">%s</span></p>' % e(push['message'].splitlines()[0]),
diff --git a/contrib/ci/lambda_functions/ci.py b/contrib/ci/lambda_functions/ci.py
--- a/contrib/ci/lambda_functions/ci.py
+++ b/contrib/ci/lambda_functions/ci.py
@@ -13,8 +13,12 @@
 import os
 import time
 import urllib.request
+import uuid
 
 import boto3
+from boto3.dynamodb.conditions import (
+    Key,
+)
 
 
 def handle_poll_repo(event, context):
@@ -40,7 +44,11 @@
             continue
 
         record = record['dynamodb']['NewImage']
-        schedule_ci(record['repo_url']['S'], record['repo']['S'], record['node']['S'])
+
+        bundle_url = record['bundle_url']['S'] if 'bundle_url' in record else None
+
+        schedule_ci(record['repo_url']['S'], record['repo']['S'],
+                    record['node']['S'], bundle_url=bundle_url)
 
 
 def handle_pending_job(event, context):
@@ -99,6 +107,42 @@
     react_to_instance_state_change(job_table, instance, state)
 
 
+def handle_try_server_upload(event, context):
+    """Handler for receiving a Try Server upload request.
+
+    This is invoked by end-users to upload a bundle and trigger jobs on it.
+    """
+    # Since the function is invoked directly, any print()ed output can be
+    # seen by the end-user. So be careful about what is logged.
+    messages = []
+
+    for a in ('bundle', 'node', 'branch', 'user', 'message'):
+        if a not in event:
+            messages.append('error: missing argument "%s"' % a)
+
+    if messages:
+        return messages
+
+    s3 = boto3.resource('s3')
+    dynamodb = boto3.resource('dynamodb')
+
+    bundle_bucket = s3.Bucket(os.environ['S3_TRY_BUNDLES_BUCKET'])
+    push_table = dynamodb.Table(os.environ['DYNAMODB_PUSH_TABLE'])
+
+    return process_try_bundle(
+        bundle_bucket,
+        push_table,
+        base64.b64decode(event['bundle']),
+        event['node'],
+        event['branch'],
+        event['user'],
+        event['message'],
+        os.environ['WEB_URL'],
+    )
+
+    return messages
+
+
 def next_build_number(job_table, repo, node, job_name):
     """Find the next available build number for a job given its unique name."""
 
@@ -190,8 +234,9 @@
     poll_table.put_item(Item=new_state)
 
 
-def schedule_ci(repo_url, repo, node):
-    print('scheduling CI for revision %s on %s' % (node, repo_url))
+def schedule_ci(repo_url, repo, node, bundle_url=None):
+    """Schedule CI from a record in the push table."""
+    print('scheduling CI for revision %s on %s' % (node, repo))
     dynamodb = boto3.resource('dynamodb')
     ec2 = boto3.resource('ec2')
     s3 = boto3.resource('s3')
@@ -203,15 +248,25 @@
 
     # TODO we should build AMIs using in-repo code so all jobs are using an
     # appropriate AMI for the revision.
+    # TODO we should store metadata in the push record that allows specifying
+    # which jobs to run.
     for image in ec2.images.filter(Owners=['self']):
         if image.name == 'hg-linux-dev-debian9':
-            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo, node, image, 'debian9')
+            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo,
+                              node, image, 'debian9',
+                              bundle_url=bundle_url)
         elif image.name == 'hg-linux-dev-debian10':
-            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo, node, image, 'debian10')
+            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo,
+                              node, image, 'debian10',
+                              bundle_url=bundle_url)
         elif image.name == 'hg-linux-dev-ubuntu18.04':
-            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo, node, image, 'ubuntu18.04')
+            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo,
+                              node, image, 'ubuntu18.04',
+                              bundle_url=bundle_url)
         elif image.name == 'hg-linux-dev-ubuntu19.04':
-            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo, node, image, 'ubuntu19.04')
+            schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo,
+                              node, image, 'ubuntu19.04',
+                              bundle_url=bundle_url)
 
 
 RUN_TESTS_LINUX = '''
@@ -219,20 +274,38 @@
 
 HG=/hgdev/venv-bootstrap/bin/hg
 
+REPO_URL=$1
+NODE=$2
+PYTHON=$3
+S3_URL=$4
+BUNDLE_URL=$5
+
 cd /hgwork/src
 
-${HG} pull -r $2 $1
-${HG} log -r $2
-${HG} up $2
+if [ -z "${BUNDLE_URL}" ]; then
+  echo "pulling $NODE from $REPO_URL"
+  ${HG} pull -r $NODE $REPO_URL
+else
+  echo "pulling $REPO_URL"
+  ${HG} pull $REPO_URL
+  echo "fetching bundle from $BUNDLE_URL"
+  aws s3 cp $BUNDLE_URL bundle.hg
+  echo "applying bundle"
+  ${HG} unbundle bundle.hg
+fi
+
+# Bail immediately if we can't find the requested revision.
+${HG} log -r $NODE || exit 1
+${HG} up $NODE || exit 1
 
 export TMPDIR=/hgwork/tmp
 cd tests
-time $3 ./run-tests.py --json 2>&1 | tee output.log
+time $PYTHON ./run-tests.py --json 2>&1 | tee output.log
 
-aws s3 cp --content-type text/plain --acl public-read output.log $4/output.log
+aws s3 cp --content-type text/plain --acl public-read output.log $S3_URL/output.log
 # The JSON file has a prefix to allow loading in web browsers.
 tail -c +13 report.json > report-truncated.json
-aws s3 cp --content-type application/json --acl public-read report-truncated.json $4/report.json
+aws s3 cp --content-type application/json --acl public-read report-truncated.json $S3_URL/report.json
 '''.lstrip()
 
 
@@ -248,6 +321,7 @@
 # TAG build_number {build_number}
 # TAG s3_bucket {s3_bucket}
 # TAG s3_prefix {s3_prefix}
+# TAG bundle_url {bundle_url}
 
 repo_update: false
 repo_upgrade: false
@@ -264,7 +338,7 @@
     - mkdir /hgwork/tmp
     - chown -R hg:hg /hgwork
     - sudo -u hg -g hg rsync -a /hgdev/src /hgwork/
-    - sudo -u hg -g hg /run-tests-linux {repo_url} {node} {python} s3://{s3_bucket}/{s3_prefix} 2>&1 | tee /ci.log
+    - sudo -u hg -g hg /run-tests-linux {repo_url} {node} {python} s3://{s3_bucket}/{s3_prefix} '{bundle_url}' 2>&1 | tee /ci.log
     - aws s3 cp --content-type text/plain --acl public-read /ci.log s3://{s3_bucket}/{s3_prefix}/ci.log
     - echo done > done
     - aws s3 cp --content-type text/plain --acl public-read done s3://{s3_bucket}/{s3_prefix}/done
@@ -276,7 +350,8 @@
 '''.lstrip()
 
 
-def schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo, node, image, os_prefix):
+def schedule_linux_ci(job_table, sqs, sqs_url, bucket, repo_url, repo,
+                      node, image, os_prefix, bundle_url=None):
     block_device_mappings = [
         {
             'DeviceName': image.block_device_mappings[0]['DeviceName'],
@@ -322,6 +397,7 @@
             run_tests_linux_b64=run_tests_linux_b64,
             s3_bucket=bucket.name,
             s3_prefix=bucket_prefix,
+            bundle_url=bundle_url or '',
         )
 
         config = {
@@ -403,7 +479,14 @@
                 continue
 
             kv = line[len('# TAG '):].strip()
-            name, value = kv.split(' ', 1)
+            parts = kv.split(' ', 1)
+
+            # Ignore empty values.
+            if len(parts) == 1:
+                continue
+
+            name, value = parts
+
             tags[name] = value
             set_tags.append({
                 'Key': name,
@@ -567,3 +650,62 @@
                     v[kk] = decimal.Decimal(v[kk])
 
             batch.put_item(Item=v)
+
+
+def process_try_bundle(bundle_bucket, push_table, bundle, node, branch, user,
+                       message, web_url):
+    """Process an incoming Try bundle."""
+    messages = []
+
+    now = datetime.datetime.utcnow().isoformat()
+
+    messages.append('processing bundle of size %d bytes' % len(bundle))
+
+    # Verify we don't already have a Try push for this node. We may need
+    # to reconsider this logic once we can cherry-pick which jobs get
+    # scheduled...
+    res = push_table.query(
+        IndexName='repo-node',
+        KeyConditionExpression=Key('repo').eq('try') & Key('node').eq(node),
+        Select='ALL_PROJECTED_ATTRIBUTES',
+    )
+    if res.get('Items'):
+        messages.append('%s previously submitted; did you forget to amend?' % node)
+        messages.append('rejecting Try submission')
+        return messages
+
+    key = 'bundle/%s.hg' % uuid.uuid4()
+
+    messages.append('uploading bundle to S3: %s' % key)
+    bundle_bucket.put_object(
+        Key=key,
+        Body=bundle,
+        ContentType='application/octet-stream',
+        ACL='private',
+        StorageClass='STANDARD_IA',
+    )
+
+    bundle_url = 's3://%s/%s' % (bundle_bucket.name, key)
+
+    # We record this as a "push" in DynamoDB. The creation of this entry
+    # will result in CI being scheduled.
+    messages.append('recording push in DynamoDB')
+    push_table.put_item(Item={
+        'repo': 'try',
+        'push_id': '%s-try' % now,
+        # Try pushes are always based on the published repository history.
+        'repo_url': 'https://www.mercurial-scm.org/repo/hg',
+        'bundle_url': bundle_url,
+        'repo_name': 'try',
+        'poll_rev': 'try',
+        'push_date': now,
+        'node': node,
+        'branch': branch,
+        'user': user,
+        'message': message,
+     })
+
+    messages.append('push recorded; jobs should start running automatically')
+    messages.append('see results at %s' % web_url)
+
+    return messages
diff --git a/contrib/ci/hgci/try_server.py b/contrib/ci/hgci/try_server.py
new file mode 100644
--- /dev/null
+++ b/contrib/ci/hgci/try_server.py
@@ -0,0 +1,93 @@
+# try_server.py - Interact with Try server
+#
+# Copyright 2019 Gregory Szorc <gregory.szorc at gmail.com>
+#
+# This software may be used and distributed according to the terms of the
+# GNU General Public License version 2 or any later version.
+
+# no-check-code because Python 3 native.
+
+import base64
+import json
+import os
+import subprocess
+import tempfile
+
+import boto3
+
+LAMBDA_FUNCTION = 'ci-try-server-upload'
+
+
+def trigger_try(rev='.'):
+    """Trigger a new Try run."""
+    lambda_client = boto3.client('lambda', region_name='us-west-2')
+
+    cset, bundle = generate_bundle(rev=rev)
+
+    payload = {
+        'bundle': base64.b64encode(bundle).decode('utf-8'),
+        'node': cset['node'],
+        'branch': cset['branch'],
+        'user': cset['user'],
+        'message': cset['desc'],
+    }
+
+    print('resolved revision:')
+    print('node: %s' % cset['node'])
+    print('branch: %s' % cset['branch'])
+    print('user: %s' % cset['user'])
+    print('desc: %s' % cset['desc'].splitlines()[0])
+    print()
+
+    print('sending to Try...')
+    res = lambda_client.invoke(
+        FunctionName=LAMBDA_FUNCTION,
+        InvocationType='RequestResponse',
+        Payload=json.dumps(payload).encode('utf-8'),
+    )
+
+    body = json.load(res['Payload'])
+    for message in body:
+        print('remote: %s' % message)
+
+
+def generate_bundle(rev='.'):
+    """Generate a bundle suitable for use by the Try service.
+
+    Returns a tuple of revision metadata and raw Mercurial bundle data.
+    """
+    # `hg bundle` doesn't support streaming to stdout. So we use a temporary
+    # file.
+    path = None
+    try:
+        fd, path = tempfile.mkstemp(prefix='hg-bundle-', suffix='.hg')
+        os.close(fd)
+
+        args = [
+            'hg', 'bundle',
+            '--type', 'gzip-v2',
+            '--base', 'public()',
+            '--rev', rev,
+            path,
+        ]
+
+        print('generating bundle...')
+        subprocess.run(args, check=True)
+
+        with open(path, 'rb') as fh:
+            bundle_data = fh.read()
+
+    finally:
+        if path:
+            os.unlink(path)
+
+    args = [
+        'hg', 'log',
+        '-r', rev,
+        # We have to upload as JSON, so it won't matter if we emit binary
+        # since we need to normalize to UTF-8.
+        '-T', 'json',
+    ]
+    res = subprocess.run(args, check=True, capture_output=True)
+    return json.loads(res.stdout)[0], bundle_data
+
diff --git a/contrib/ci/hgci/cli.py b/contrib/ci/hgci/cli.py
new file mode 100644
--- /dev/null
+++ b/contrib/ci/hgci/cli.py
@@ -0,0 +1,53 @@
+# cli.py - Command line interface for CI
+#
+# Copyright 2019 Gregory Szorc <gregory.szorc at gmail.com>
+#
+# This software may be used and distributed according to the terms of the
+# GNU General Public License version 2 or any later version.
+
+# no-check-code because Python 3 native.
+
+import argparse
+import os
+import pathlib
+
+from . import (
+    try_server,
+)
+
+SOURCE_ROOT = pathlib.Path(os.path.abspath(__file__)).parent.parent.parent.parent
+
+
+def run_try(rev):
+    try_server.trigger_try(rev=rev)
+
+
+def get_parser():
+    parser = argparse.ArgumentParser()
+
+    subparsers = parser.add_subparsers()
+
+    sp = subparsers.add_parser(
+        'try',
+        help='Run CI automation against a custom changeset'
+    )
+    sp.add_argument('-r', '--rev',
+                    default='.',
+                    help='Revision to run CI on')
+    sp.set_defaults(func=run_try)
+
+    return parser
+
+
+def main():
+    parser = get_parser()
+    args = parser.parse_args()
+
+    if not hasattr(args, 'func'):
+        parser.print_help()
+        return
+
+    kwargs = dict(vars(args))
+    del kwargs['func']
+
+    args.func(**kwargs)
diff --git a/contrib/ci/hgci/__init__.py b/contrib/ci/hgci/__init__.py
new file mode 100644
diff --git a/contrib/ci/ci.py b/contrib/ci/ci.py
new file mode 100755
--- /dev/null
+++ b/contrib/ci/ci.py
@@ -0,0 +1,68 @@
+#!/usr/bin/env python3
+#
+# ci.py - Interact with Mercurial CI
+#
+# Copyright 2019 Gregory Szorc <gregory.szorc at gmail.com>
+#
+# This software may be used and distributed according to the terms of the
+# GNU General Public License version 2 or any later version.
+
+import os
+import pathlib
+import subprocess
+import sys
+import venv
+
+
+HERE = pathlib.Path(os.path.abspath(__file__)).parent
+REQUIREMENTS_TXT = HERE / 'requirements.txt'
+SOURCE_DIR = HERE.parent.parent
+VENV = SOURCE_DIR / 'build' / 'venv-ci'
+
+
+def bootstrap():
+    venv_created = not VENV.exists()
+
+    VENV.parent.mkdir(exist_ok=True)
+
+    venv.create(VENV, with_pip=True)
+
+    if os.name == 'nt':
+        venv_bin = VENV / 'Scripts'
+        pip = venv_bin / 'pip.exe'
+        python = venv_bin / 'python.exe'
+    else:
+        venv_bin = VENV / 'bin'
+        pip = venv_bin / 'pip'
+        python = venv_bin / 'python'
+
+    args = [str(pip), 'install', '-r', str(REQUIREMENTS_TXT),
+            '--disable-pip-version-check']
+
+    if not venv_created:
+        args.append('-q')
+
+    subprocess.run(args, check=True)
+
+    os.environ['HGCI_BOOTSTRAPPED'] = '1'
+    os.environ['PATH'] = '%s%s%s' % (
+        venv_bin, os.pathsep, os.environ['PATH'])
+
+    subprocess.run([str(python), __file__] + sys.argv[1:], check=True)
+
+
+def run():
+    import hgci.cli as cli
+    cli.main()
+
+
+if __name__ == '__main__':
+    try:
+        if 'HGCI_BOOTSTRAPPED' not in os.environ:
+            bootstrap()
+        else:
+            run()
+    except subprocess.CalledProcessError as e:
+        sys.exit(e.returncode)
+    except KeyboardInterrupt:
+        sys.exit(1)
diff --git a/contrib/ci/README.rst b/contrib/ci/README.rst
--- a/contrib/ci/README.rst
+++ b/contrib/ci/README.rst
@@ -4,6 +4,54 @@
 
 This directory defines a CI system for the Mercurial Project.
 
+Try Server
+==========
+
+The CI system features a *Try Server* which allows you to submit
+arbitrary changesets for evaluation. This works by sending a
+Mercurial bundle to a remote server, which ingests it into the CI
+system and schedules it for execution.
+
+Registering
+-----------
+
+Because the *Try Server* is remote code execution as a service,
+we require users to register to use it and all requests must be
+authenticated.
+
+To register a new account, you will need to add the requested user
+to the ``terraform/iam_users.tf`` file. Create a new
+``aws_iam_user`` resource representing the new user and add this
+user to the ``access_to_try_upload`` ``aws_iam_group_membership``
+resource also in that file.
+
+This change will need to be submitted for review and applied by an
+admin of the CI system.
+
+TODO document how to configure authentication credentials
+
+Using
+-----
+
+Once you have your AWS credentials configured, using the Try server
+is simple::
+
+   $ contrib/ci/ci.py try
+   generating bundle...
+   37 changesets found
+   resolved revision:
+   node: 675ee4827c97bc85ae519b37a5be6ebc95c17b80
+   branch: default
+   user: Gregory Szorc <gregory.szorc at gmail.com>
+   desc: ci: implement a "try server"
+
+   sending to Try...
+   remote: processing bundle of size 49900 bytes
+   remote: uploading bundle to S3: bundle/cf667fe0-f8ce-44db-978e-7fd36756aa86.hg
+   remote: recording push in DynamoDB
+   remote: push recorded; jobs should start running automatically
+   remote: see results at https://ci.hg.gregoryszorc.com/
+
 Architecture
 ============
 
@@ -126,6 +174,20 @@
 
 The Terraform code for this component lives in ``web.tf``.
 
+Try Server
+----------
+
+The *try server* components provides a mechanism to trigger CI on arbitrary
+diffs. It allows Mercurial developers to test ad-hoc changes by uploading
+Mercurial bundles to the CI system.
+
+The *try server* works by exposing a Lambda function to privileged
+users. The Lambda function is invoked with a Mercurial bundle and metadata
+describing the requested jobs. The function then stores the bundle in an
+S3 bucket and registers jobs to run against the uploaded bundle.
+
+The Terraform code for this component lives in ``try_server.tf``.
+
 AWS Account Management
 ======================
 
@@ -204,13 +266,14 @@
 We have no mechanism to retrigger a job. This requires some form of
 authentication to prevent abuse.
 
-We have no mechanism to trigger CI on arbitrary diffs. We would like
-to provide some kind of *try server* where you can submit a diff and
-the system builds it. Again, this requires some form of authentication.
-
 We have no mechanism to choose which jobs to execute. We probably want
 to build this because there is no need to execute all jobs all the time.
 
+The association between jobs, job results, and pushes is a bit wonky.
+Things can get confused if there are multiple pushes for the same
+changeset. We may need to change the primary key of the DynamoDB tables
+to fix this.
+
 Development Workflow
 ====================
 



To: indygreg, #hg-reviewers
Cc: mercurial-devel


More information about the Mercurial-devel mailing list