Merge pull request #170 from jonas-lq/issue#109

Add bot comments in app.cfg.example

Looks good. Tested with trz42/software-layer#61

Minor polishing which can be done in a follow-up PR.

app.cfg.example

@@ -138,3 +138,24 @@ poll_interval = 60
 
 # full path to the command for manipulating existing jobs
 scontrol_command = /usr/bin/scontrol
+
+# variable 'comment' under 'submitted_job_comments' should not be changed as there are regular expression patterns matching it
+[submitted_job_comments]
+initial_comment = New job on instance `{app_name}` for architecture `{arch_name}` for repository `{repo_id}` in job dir `{symlink}` 
+awaits_release = job id `{job_id}` awaits release by job manager 
+
+[new_job_comments]
+awaits_lauch = job awaits launch by Slurm scheduler 
+
+[running_job_comments]
+running_job = job `{job_id}` is running 
+
+[finished_job_comments]
+success = :grin: SUCCESS tarball `{tarball_name}` ({tarball_size} GiB) in job dir 
+failure = :cry: FAILURE 
+no_slurm_out = No slurm output `{slurm_out}` in job dir 
+slurm_out = Found slurm output `{slurm_out}` in job dir 
+missing_modules = Slurm output lacks message "No missing modules!". 
+no_tarball_message = Slurm output lacks message about created tarball. 
+no_matching_tarball = No tarball matching `{tarball_pattern}` found in job dir. 
+multiple_tarballs = Found {num_tarballs} tarballs in job dir - only 1 matching `{tarball_pattern}` expected. 

eessi_bot_event_handler.py

@@ -15,6 +15,7 @@
 #
 import waitress
 import sys
+import tasks.build as build
 
 from connections import github
 from tools import config
@@ -162,8 +163,11 @@ def main():
     """Main function."""
     opts = event_handler_parse()
 
-    # config is read to raise an exception early when the event_handler starts.
-    config.read_config()
+    required_config = {
+        build.SUBMITTED_JOB_COMMENTS: [build.INITIAL_COMMENT, build.AWAITS_RELEASE]
+    }
+    # config is read and checked for settings to raise an exception early when the event_handler starts.
+    config.check_required_cfg_settings(required_config)
     github.connect()
 
     if opts.file:

eessi_bot_job_manager.py

@@ -43,6 +43,27 @@
 
 from pyghee.utils import log, error
 
+AWAITS_LAUCH = "awaits_lauch"
+FAILURE = "failure"
+FINISHED_JOB_COMMENTS = "finished_job_comments"
+NEW_JOB_COMMENTS = "new_job_comments"
+MISSING_MODULES = "missing_modules"
+MULTIPLE_TARBALLS = "multiple_tarballs"
+NO_MATCHING_TARBALL = "no_matching_tarball"
+NO_SLURM_OUT = "no_slurm_out"
+NO_TARBALL_MESSAGE = "no_tarball_message"
+RUNNING_JOB = "running_job"
+RUNNING_JOB_COMMENTS = "running_job_comments"
+SLURM_OUT = "slurm_out"
+SUCCESS = "success"
+
+REQUIRED_CONFIG = {
+    NEW_JOB_COMMENTS: [AWAITS_LAUCH],
+    RUNNING_JOB_COMMENTS: [RUNNING_JOB],
+    FINISHED_JOB_COMMENTS: [SUCCESS, FAILURE, NO_SLURM_OUT, SLURM_OUT, MISSING_MODULES,
+                            NO_TARBALL_MESSAGE, NO_MATCHING_TARBALL, MULTIPLE_TARBALLS]
+}
+
 
 class EESSIBotSoftwareLayerJobManager:
     "main class for (Slurm) job manager of EESSI bot (separate process)"
@@ -291,9 +312,10 @@ def process_new_job(self, new_job):
             # (c) add a row to the table
             # add row to status table if we found a comment
             if "comment_id" in new_job:
+                new_job_comments_cfg = config.read_config()[NEW_JOB_COMMENTS]
                 dt = datetime.now(timezone.utc)
-                update = "\n|%s|" % dt.strftime("%b %d %X %Z %Y")
-                update += "released|job awaits launch by Slurm scheduler|"
+                update = "\n|%s|released|" % dt.strftime("%b %d %X %Z %Y")
+                update += f"{new_job_comments_cfg[AWAITS_LAUCH]}|"
                 update_comment(new_job["comment_id"], pr, update)
             else:
                 log(
@@ -363,12 +385,13 @@ def process_running_jobs(self, running_job):
 
         if "comment_id" in running_job:
             dt = datetime.now(timezone.utc)
-            running_msg = "job %s is running" % running_job['jobid']
+            running_job_comments_cfg = config.read_config()[RUNNING_JOB_COMMENTS]
+            running_msg = running_job_comments_cfg[RUNNING_JOB].format(job_id=running_job['jobid'])
             if "comment_body" in running_job and running_msg in running_job["comment_body"]:
                 log("Not updating comment, '%s' already found" % running_msg)
             else:
-                update = "\n|%s|" % dt.strftime("%b %d %X %Z %Y")
-                update += "running|" + running_msg
+                update = f"\n|{dt.strftime('%b %d %X %Z %Y')}|running|"
+                update += f"{running_msg}|"
                 update_comment(running_job["comment_id"], pullrequest, update)
         else:
             log(
@@ -463,20 +486,20 @@ def process_finished_job(self, finished_job):
 
         dt = datetime.now(timezone.utc)
 
+        finished_job_comments_cfg = config.read_config()[FINISHED_JOB_COMMENTS]
+        comment_update = f"\n|{dt.strftime('%b %d %X %Z %Y')}|finished|"
         if (no_missing_modules and targz_created and
                 len(eessi_tarballs) == 1):
             # We've got one tarball and slurm out messages are ok
             # Prepare a message with information such as
             #   (installation status, tarball name, tarball size)
-            comment_update = "\n|%s|finished|:grin: SUCCESS " % dt.strftime(
-                 "%b %d %X %Z %Y")
-
+            tarball_name = os.path.basename(eessi_tarballs[0])
             tarball_size = os.path.getsize(eessi_tarballs[0]) / 2**30
-            comment_update += "tarball <code>%s</code> (%.3f GiB) " % (
-                os.path.basename(eessi_tarballs[0]),
-                tarball_size,
+            success_comment = finished_job_comments_cfg[SUCCESS].format(
+                tarball_name=tarball_name,
+                tarball_size=tarball_size
             )
-            comment_update += "in job dir|"
+            comment_update += f"{success_comment}|"
             # NOTE explicitly name repo in build job comment?
             # comment_update += '\nAwaiting approval to
             #  comment_update +=  ingest tarball into the repository.'
@@ -489,52 +512,39 @@ def process_finished_job(self, finished_job):
             # prepare a message with details about the above conditions and
             # update PR with a comment
 
-            comment_update = "\n|%s|finished|:cry: FAILURE <ul>" % dt.strftime(
-                "%b %d %X %Z %Y"
-            )
+            comment_update += f"{finished_job_comments_cfg[FAILURE]} <ul>"
             found_slurm_out = os.path.exists(slurm_out)
 
             if not found_slurm_out:
-                # no slurm out ... something went wrong with the job
-                comment_update += (
-                    "<li>No slurm output <code>%s</code> in job dir</li>"
-                    % os.path.basename(slurm_out)
+                # no slurm out ... something went wrong with the job f"<li> {} </li>"
+                comment_update += f"<li> {finished_job_comments_cfg[NO_SLURM_OUT]} </li>".format(
+                    slurm_out=os.path.basename(slurm_out)
                 )
             else:
-                comment_update += (
-                    "<li>Found slurm output <code>%s</code> in job dir</li>"
-                    % os.path.basename(slurm_out)
+                comment_update += f"<li> {finished_job_comments_cfg[SLURM_OUT]} </li>".format(
+                    slurm_out=os.path.basename(slurm_out)
                 )
 
             if found_slurm_out and not no_missing_modules:
                 # Found slurm out, but doesn't contain message 'No missing modules!'
-                comment_update += (
-                    '<li>Slurm output lacks message "No missing modules!".</li>'
-                )
+                comment_update += f"<li> {finished_job_comments_cfg[MISSING_MODULES]} </li>"
 
             if found_slurm_out and not targz_created:
                 # Found slurm out, but doesn't contain message
                 #   'eessi-.*-software-.*.tar.gz created!'
-                comment_update += (
-                    "<li>Slurm output lacks message about created tarball.</li>"
-                )
+                comment_update += f"<li> {finished_job_comments_cfg[NO_TARBALL_MESSAGE]} </li>"
 
             if len(eessi_tarballs) == 0:
                 # no luck, job just seemed to have failed ...
-                comment_update += (
-                    "<li>No tarball matching <code>%s</code> found in job dir.</li>"
-                    % tarball_pattern.replace(r"*", r"\*")
+                comment_update += f"<li> {finished_job_comments_cfg[NO_MATCHING_TARBALL]} </li>".format(
+                    tarball_pattern=tarball_pattern.replace(r"*", r"\*")
                 )
 
             if len(eessi_tarballs) > 1:
                 # something's fishy, we only expected a single tar.gz file
-                comment_update += (
-                    "<li>Found %d tarballs in job dir - only 1 "
-                    "matching <code>%s</code> expected.</li>"
-                    % (
-                        len(eessi_tarballs),
-                        tarball_pattern.replace(r"*", r"\*"),
-                    )
+                comment_update += f"<li> {finished_job_comments_cfg[MULTIPLE_TARBALLS]} </li>".format(
+                    num_tarballs=len(eessi_tarballs),
+                    tarball_pattern=tarball_pattern.replace(r"*", r"\*")
                 )
             comment_update += "</ul>|"
             # comment_update += '\nAn admin may investigate what went wrong.
@@ -572,8 +582,8 @@ def main():
 
     opts = job_manager_parse()
 
-    # config is read to raise an exception early when the job_manager runs.
-    config.read_config()
+    # config is read and checked for settings to raise an exception early when the job_manager runs.
+    config.check_required_cfg_settings(REQUIRED_CONFIG)
     github.connect()
 
     job_manager = EESSIBotSoftwareLayerJobManager()

tasks/build.py

@@ -23,17 +23,20 @@
 from pyghee.utils import log, error
 from tools import config, run_cmd
 
+AWAITS_RELEASE = "awaits_release"
 BUILDENV = "buildenv"
 BUILD_JOB_SCRIPT = "build_job_script"
 CONTAINER_CACHEDIR = "container_cachedir"
 DEFAULT_JOB_TIME_LIMIT = "24:00:00"
 CVMFS_CUSTOMIZATIONS = "cvmfs_customizations"
 HTTP_PROXY = "http_proxy"
 HTTPS_PROXY = "https_proxy"
+INITIAL_COMMENT = "initial_comment"
 JOBS_BASE_DIR = "jobs_base_dir"
 LOAD_MODULES = "load_modules"
 LOCAL_TMP = "local_tmp"
 SLURM_PARAMS = "slurm_params"
+SUBMITTED_JOB_COMMENTS = "submitted_job_comments"
 SUBMIT_COMMAND = "submit_command"
 BUILD_PERMISSION = "build_permission"
 ARCHITECTURE_TARGETS = "architecturetargets"
@@ -552,14 +555,17 @@ def create_pr_comments(job, job_id, app_name, job_comment, pr, repo_name, gh, sy
     dt = datetime.now(timezone.utc)
 
     # construct initial job comment
-    job_comment = (f"New job on instance `{app_name}`"
-                   f" for architecture `{arch_name}`"
-                   f" for repository `{job.repo_id}`"
-                   f" in job dir `{symlink}`\n"
-                   f"|date|job status|comment|\n"
+    submitted_job_comments_cfg = config.read_config()[SUBMITTED_JOB_COMMENTS]
+    job_comment = (f"{submitted_job_comments_cfg[INITIAL_COMMENT]}"
+                   f"\n|date|job status|comment|\n"
                    f"|----------|----------|------------------------|\n"
-                   f"|{dt.strftime('%b %d %X %Z %Y')}|submitted|"
-                   f"job id `{job_id}` awaits release by job manager|")
+                   f"|{dt.strftime('%b %d %X %Z %Y')}|"
+                   f"submitted|"
+                   f"{submitted_job_comments_cfg[AWAITS_RELEASE]}|").format(app_name=app_name,
+                                                                            arch_name=arch_name,
+                                                                            symlink=symlink,
+                                                                            repo_id=job.repo_id,
+                                                                            job_id=job_id)
 
     # create comment to pull request
     repo = gh.get_repo(repo_name)

tools/config.py

@@ -30,3 +30,21 @@ def read_config(path='app.cfg'):
         error(f"{fn}(): Unable to read configuration file {path}!\n{err}")
 
     return config
+
+
+def check_required_cfg_settings(req_settings, path="app.cfg"):
+    """
+    Reads the config file and checks if it contains the required settings, signaling an error if not
+    Args:
+        req_settings (dict (str, list)): required settings
+        path (string): path to the configuration file
+    Returns:
+        None
+    """
+    cfg = read_config()
+    for section in req_settings.keys():
+        if section not in cfg:
+            error(f'Missing section "{section}" in configuration file {path}.')
+        for item in req_settings[section]:
+            if item not in cfg[section]:
+                error(f'Missing configuration item "{item}" in section "{section}" of configuration file {path}.')