import json
import re
from functools import partial
+import subprocess
import time
import urllib
import cwltool.workflow
from schema_salad.sourceline import SourceLine
import schema_salad.validate as validate
+from schema_salad.ref_resolver import file_uri, uri_file_path
import arvados
import arvados.config
kind = 'error'
elif record.levelno >= logging.WARNING:
kind = 'warning'
+ if kind == 'warning' and record.name == "salad":
+ # Don't send validation warnings to runtime status,
+ # they're noisy and unhelpful.
+ return
if kind is not None and self.updatingRuntimeStatus is not True:
self.updatingRuntimeStatus = True
try:
arvargs.output_tags = None
arvargs.thread_count = 1
arvargs.collection_cache_size = None
+ arvargs.git_info = True
+ arvargs.submit = False
+ arvargs.defer_downloads = False
self.api = api_client
self.processes = {}
self.fs_access = None
self.secret_store = None
self.stdout = stdout
+ self.fast_submit = False
+ self.git_info = arvargs.git_info
if keep_client is not None:
self.keep_client = keep_client
self.toplevel_runtimeContext.make_fs_access = partial(CollectionFsAccess,
collection_cache=self.collection_cache)
+ self.defer_downloads = arvargs.submit and arvargs.defer_downloads
+
validate_cluster_target(self, self.toplevel_runtimeContext)
Called when there's a need to report errors, warnings or just
activity statuses, for example in the RuntimeStatusLoggingHandler.
"""
+
+ if kind not in ('error', 'warning'):
+ # Ignore any other status kind
+ return
+
with self.workflow_eval_lock:
current = None
try:
if current is None:
return
runtime_status = current.get('runtime_status', {})
- if kind in ('error', 'warning'):
- updatemessage = runtime_status.get(kind, "")
- if not updatemessage:
- updatemessage = message
-
- # Subsequent messages tacked on in detail
- updatedetail = runtime_status.get(kind+'Detail', "")
- maxlines = 40
- if updatedetail.count("\n") < maxlines:
- if updatedetail:
- updatedetail += "\n"
- updatedetail += message + "\n"
-
- if detail:
- updatedetail += detail + "\n"
-
- if updatedetail.count("\n") >= maxlines:
- updatedetail += "\nSome messages may have been omitted. Check the full log."
-
- runtime_status.update({
- kind: updatemessage,
- kind+'Detail': updatedetail,
- })
- else:
- # Ignore any other status kind
+
+ original_updatemessage = updatemessage = runtime_status.get(kind, "")
+ if not updatemessage:
+ updatemessage = message
+
+ # Subsequent messages tacked on in detail
+ original_updatedetail = updatedetail = runtime_status.get(kind+'Detail', "")
+ maxlines = 40
+ if updatedetail.count("\n") < maxlines:
+ if updatedetail:
+ updatedetail += "\n"
+ updatedetail += message + "\n"
+
+ if detail:
+ updatedetail += detail + "\n"
+
+ if updatedetail.count("\n") >= maxlines:
+ updatedetail += "\nSome messages may have been omitted. Check the full log."
+
+ if updatemessage == original_updatemessage and updatedetail == original_updatedetail:
+ # don't waste time doing an update if nothing changed
+ # (usually because we exceeded the max lines)
return
+
+ runtime_status.update({
+ kind: updatemessage,
+ kind+'Detail': updatedetail,
+ })
+
try:
self.api.containers().update(uuid=current['uuid'],
body={
page = keys[:pageSize]
try:
proc_states = table.list(filters=[["uuid", "in", page]]).execute(num_retries=self.num_retries)
- except Exception:
- logger.exception("Error checking states on API server: %s")
+ except Exception as e:
+ logger.exception("Error checking states on API server: %s", e)
remain_wait = self.poll_interval
continue
for req in job_reqs:
tool.requirements.append(req)
+ @staticmethod
+ def get_git_info(tool):
+ in_a_git_repo = False
+ cwd = None
+ filepath = None
+
+ if tool.tool["id"].startswith("file://"):
+ # check if git is installed
+ try:
+ filepath = uri_file_path(tool.tool["id"])
+ cwd = os.path.dirname(filepath)
+ subprocess.run(["git", "log", "--format=%H", "-n1", "HEAD"], cwd=cwd, check=True, capture_output=True, text=True)
+ in_a_git_repo = True
+ except Exception as e:
+ pass
+
+ gitproperties = {}
+
+ if in_a_git_repo:
+ git_commit = subprocess.run(["git", "log", "--format=%H", "-n1", "HEAD"], cwd=cwd, capture_output=True, text=True).stdout
+ git_date = subprocess.run(["git", "log", "--format=%cD", "-n1", "HEAD"], cwd=cwd, capture_output=True, text=True).stdout
+ git_committer = subprocess.run(["git", "log", "--format=%cn <%ce>", "-n1", "HEAD"], cwd=cwd, capture_output=True, text=True).stdout
+ git_branch = subprocess.run(["git", "rev-parse", "--abbrev-ref", "HEAD"], cwd=cwd, capture_output=True, text=True).stdout
+ git_origin = subprocess.run(["git", "remote", "get-url", "origin"], cwd=cwd, capture_output=True, text=True).stdout
+ git_status = subprocess.run(["git", "status", "--untracked-files=no", "--porcelain"], cwd=cwd, capture_output=True, text=True).stdout
+ git_describe = subprocess.run(["git", "describe", "--always", "--tags"], cwd=cwd, capture_output=True, text=True).stdout
+ git_toplevel = subprocess.run(["git", "rev-parse", "--show-toplevel"], cwd=cwd, capture_output=True, text=True).stdout
+ git_path = filepath[len(git_toplevel):]
+
+ gitproperties = {
+ "http://arvados.org/cwl#gitCommit": git_commit.strip(),
+ "http://arvados.org/cwl#gitDate": git_date.strip(),
+ "http://arvados.org/cwl#gitCommitter": git_committer.strip(),
+ "http://arvados.org/cwl#gitBranch": git_branch.strip(),
+ "http://arvados.org/cwl#gitOrigin": git_origin.strip(),
+ "http://arvados.org/cwl#gitStatus": git_status.strip(),
+ "http://arvados.org/cwl#gitDescribe": git_describe.strip(),
+ "http://arvados.org/cwl#gitPath": git_path.strip(),
+ }
+ else:
+ for g in ("http://arvados.org/cwl#gitCommit",
+ "http://arvados.org/cwl#gitDate",
+ "http://arvados.org/cwl#gitCommitter",
+ "http://arvados.org/cwl#gitBranch",
+ "http://arvados.org/cwl#gitOrigin",
+ "http://arvados.org/cwl#gitStatus",
+ "http://arvados.org/cwl#gitDescribe",
+ "http://arvados.org/cwl#gitPath"):
+ if g in tool.metadata:
+ gitproperties[g] = tool.metadata[g]
+
+ return gitproperties
+
+ def set_container_request_properties(self, container, properties):
+ resp = self.api.container_requests().list(filters=[["container_uuid", "=", container["uuid"]]], select=["uuid", "properties"]).execute(num_retries=self.num_retries)
+ for cr in resp["items"]:
+ cr["properties"].update({k.replace("http://arvados.org/cwl#", "arv:"): v for k, v in properties.items()})
+ self.api.container_requests().update(uuid=cr["uuid"], body={"container_request": {"properties": cr["properties"]}}).execute(num_retries=self.num_retries)
+
def arv_executor(self, updated_tool, job_order, runtimeContext, logger=None):
self.debug = runtimeContext.debug
+ git_info = self.get_git_info(updated_tool) if self.git_info else {}
+ if git_info:
+ logger.info("Git provenance")
+ for g in git_info:
+ if git_info[g]:
+ logger.info(" %s: %s", g.split("#", 1)[1], git_info[g])
+
workbench1 = self.api.config()["Services"]["Workbench1"]["ExternalURL"]
workbench2 = self.api.config()["Services"]["Workbench2"]["ExternalURL"]
controller = self.api.config()["Services"]["Controller"]["ExternalURL"]
logger.info("Using cluster %s (%s)", self.api.config()["ClusterID"], workbench2 or workbench1 or controller)
- updated_tool.visit(self.check_features)
+ if not self.fast_submit:
+ updated_tool.visit(self.check_features)
self.pipeline = None
self.fs_access = runtimeContext.make_fs_access(runtimeContext.basedir)
runtimeContext.intermediate_storage_classes = default_storage_classes
if not runtimeContext.name:
- runtimeContext.name = self.name = updated_tool.tool.get("label") or updated_tool.metadata.get("label") or os.path.basename(updated_tool.tool["id"])
+ self.name = updated_tool.tool.get("label") or updated_tool.metadata.get("label") or os.path.basename(updated_tool.tool["id"])
+ if git_info.get("http://arvados.org/cwl#gitDescribe"):
+ self.name = "%s (%s)" % (self.name, git_info.get("http://arvados.org/cwl#gitDescribe"))
+ runtimeContext.name = self.name
if runtimeContext.copy_deps is None and (runtimeContext.create_workflow or runtimeContext.update_workflow):
# When creating or updating workflow record, by default
loadingContext = self.loadingContext.copy()
loadingContext.do_validate = False
loadingContext.disable_js_validation = True
- if submitting:
+ if submitting and not self.fast_submit:
loadingContext.do_update = False
# Document may have been auto-updated. Reload the original
# document with updating disabled because we want to
# Upload direct dependencies of workflow steps, get back mapping of files to keep references.
# Also uploads docker images.
- logger.info("Uploading workflow dependencies")
- with Perf(metrics, "upload_workflow_deps"):
- merged_map = upload_workflow_deps(self, tool, runtimeContext)
+ if not self.fast_submit:
+ logger.info("Uploading workflow dependencies")
+ with Perf(metrics, "upload_workflow_deps"):
+ merged_map = upload_workflow_deps(self, tool, runtimeContext)
+ else:
+ merged_map = {}
# Recreate process object (ArvadosWorkflow or
# ArvadosCommandTool) because tool document may have been
submit_runner_ram=runtimeContext.submit_runner_ram,
name=runtimeContext.name,
merged_map=merged_map,
- submit_runner_image=runtimeContext.submit_runner_image)
+ submit_runner_image=runtimeContext.submit_runner_image,
+ git_info=git_info)
self.stdout.write(uuid + "\n")
return (None, "success")
priority=runtimeContext.priority,
secret_store=self.secret_store,
collection_cache_size=runtimeContext.collection_cache_size,
- collection_cache_is_default=self.should_estimate_cache_size)
+ collection_cache_is_default=self.should_estimate_cache_size,
+ git_info=git_info)
else:
runtimeContext.runnerjob = tool.tool["id"]
current_container = arvados_cwl.util.get_current_container(self.api, self.num_retries, logger)
if current_container:
logger.info("Running inside container %s", current_container.get("uuid"))
+ self.set_container_request_properties(current_container, git_info)
self.poll_api = arvados.api('v1', timeout=runtimeContext.http_timeout)
self.polling_thread = threading.Thread(target=self.poll_states)