tcwg-report-stale-rr-jobs.sh: use manifest to detect unupdated branches

Change-Id: I86f1e5fae54106a0113b6714300ccbba98947bfe

1 files changed, 106 insertions, 44 deletions

diff --git a/tcwg-report-stale-rr-jobs.sh b/tcwg-report-stale-rr-jobs.sh
index 06ce4536..1dd7560b 100755
--- a/tcwg-report-stale-rr-jobs.sh
+++ b/tcwg-report-stale-rr-jobs.sh
@@ -19,22 +19,38 @@ only="${only-false}"
 keep_tmp="${keep_tmp-false}"
 tmpdir="${tmpdir-""}"
 
-if [ x"${repos[*]}" = x"default" ]; then
-    repos=(base-artifacts binutils-gdb gcc glibc linux llvm-project newlib qemu)
-fi
+process_manifest ()
+{
+    components=$*
+    iterate_again=false
+
+    declare -A rr
+    manifest=base-artifacts/manifest.sh
+    [ -f base-artifacts/jenkins/manifest.sh ] && manifest=base-artifacts/jenkins/manifest.sh
+     # shellcheck disable=SC1090
+    source $manifest
+
+    for c in $components; do
+      if [ ! -v rr[debug_${c}_date] ]; then
+	iterate_again=true
+      else
+        commit_stamps[$c]=$(eval echo "\${rr[debug_${c}_date]}")
+      fi
+    done
+    unset rr
+}
+
 
-process_git_url ()
+process_all_artifacts ()
 {
     (
     set -euf -o pipefail
-    local refs_url="$1"
+    local baseartifacts_url="$refs_url_prefix/base-artifacts"
     local days_for_master days_for_release days_limit
 
     # Initialize base-artifacts repo (by cloning its "empty" branch).
-    refs_repo=$(basename "$refs_url" .git)
-    clone_or_update_repo_no_checkout "$refs_repo" "$refs_url" auto empty origin \
-				     >/dev/null 2>&1
-    git -C "$refs_repo" reset --hard >/dev/null 2>&1
+    clone_or_update_repo "base-artifacts" empty "$baseartifacts_url.git" >/dev/null 2>&1
+    git -C "base-artifacts" reset --hard >/dev/null 2>&1
 
     days_for_master="$days"
     days_for_release="$((days+days))"
@@ -42,23 +58,58 @@ process_git_url ()
     # Walk through all commits of all tcwg_bmk* branches and mark results
     # referenced in those results with "used_by" file.
     while IFS= read -r ref; do
+
+	if [ $only != false ] && [[ ! $ref =~ $only ]]; then
+	  continue
+	fi
+
 	dst_ref="refs/remotes/origin/${ref#refs/heads/}"
-	git -C "$refs_repo" fetch -q origin "+$ref:$dst_ref" >/dev/null 2>&1
-	commit_stamp=$(git -C "$refs_repo" show --no-patch --pretty=%ct "$dst_ref")
-	days_ago=$((($(date +%s) - $commit_stamp) / (24 * 3600)))
-        if [[ $ref =~ -release- ]]; then
-            days_limit=$days_for_release
-        else
-            days_limit=$days_for_master
-        fi
-	if [ "$days_ago" -gt "$days_limit" ]; then
-	    if $human; then
-		echo "$refs_repo: ${ref#$refs_prefix}: last updated $days_ago days ago"
-	    else
-		echo "$refs_repo:$ref"
-	    fi
+	git -C "base-artifacts" fetch -q origin "+$ref:$dst_ref" >/dev/null 2>&1
+	git -C "base-artifacts" checkout FETCH_HEAD  >/dev/null 2>&1
+
+	# -- get components for this project
+	components=""
+	if [ ! -d "base-artifacts"/git ]; then
+	  # For old style artifacts, we consider, they have not been updated since
+	  # git directory has been created in base-artifacts. i.e.
+	  #   100fd17  round-robin.sh: Fix "git rev-parse" of short histories
+	  commit_stamp=1660557250
+	  days_ago=$((($(date +%s) - $commit_stamp) / (24 * 3600)))
+	  echo "all: ${ref#$refs_prefix}: No successful run since $days_ago days"
+	  continue;
 	fi
-    done < <(git ls-remote "$refs_url" "${refs_prefix}*" | awk '{ print $2 }')
+	while read -e c; do
+          c=${c#base-artifacts/git/}
+          c=${c%_rev}
+	  components="$components $c"
+	done < <(find base-artifacts/git -name '*_rev')
+
+	# -- get all component commit stamps
+	declare -A commit_stamps
+	iterate_again=true
+	while $iterate_again ; do
+	  process_manifest $components
+          if $iterate_again; then
+	    ( git -C "base-artifacts" checkout HEAD~1 >/dev/null 2>&1 ) || iterate_again=false
+	  fi
+	done
+
+	# -- dump the messages
+	days_limit=$days_for_master
+	[[ $ref =~ -release- ]] && days_limit=$days_for_release
+
+	for c in $components; do
+	  if [[ -v commit_stamps[$c] ]]; then
+	    days_ago=$((($(date +%s) - ${commit_stamps[$c]}) / (24 * 3600)))
+	    if [ "$days_ago" -gt "$days_limit" ]; then
+	      echo "$c: ${ref#$refs_prefix}: last updated $days_ago days ago"
+	    fi
+	  else
+	    echo "$c: ${ref#$refs_prefix}: no date for this component"
+	  fi
+	done
+
+    done < <(git ls-remote "$baseartifacts_url" "${refs_prefix}*" | awk '{ print $2 }')
     )
 }
 
@@ -84,8 +135,8 @@ classify_get_project()
        ['run_nb']="" ['run_date']="" ['run_status']="" ['run_title']="" ['run_check_regression']=""
        #
        ['last_run']="" ['diag']="" )
-       
-    test['last_updated']=$(echo $1 | sed -e 's|.*: last updated ||')
+
+    test['last_updated']=$(echo $1 | sed -e 's|.*: last updated ||' -e 's|.*: No successful run since ||')
     test['gitproject']=$(echo $1 |cut -d: -f 1)
     test['branch']=$(echo $1 |cut -d: -f 2)
     test['branch']=${test['branch']:1}
@@ -167,7 +218,7 @@ classify ()
     local expression="$3"
     local diag_error="$4"
 
-    # Only if not already classified 
+    # Only if not already classified
     if [ ! -z "${test['diag']}" ]; then return; fi
 
    download_project_file "$filename"
@@ -229,7 +280,7 @@ get_artifact_dir ()
 get_run_title_and_status ()
 {
      run=$1
-     
+
      verbose "  - get_run_title_and_status() : $run"
      # Last run
      download_project_file $run/__toppage__
@@ -256,7 +307,8 @@ get_last_interesting_run ()
      export r   # to avoid shellcheck unused warning
      for r in {1..8}; do
         get_run_title_and_status ${test['run_nb']}
-        if [[ "${test['run_title']}" =~ $gitprojectshort ]]; then
+        if [[ "${test['run_title']}" =~ $gitprojectshort ]] ||
+	   [ $gitprojectshort == "all" ]; then
           test['last_run']=${test['run_nb']}
           verbose "  > ${test['run_nb']}"
           return
@@ -268,7 +320,6 @@ get_last_interesting_run ()
 
 ################## CLASSIFY FUNCTIONS
 
-
 classify_polling_error ()
 {
      if [ ! -z "${test['diag']}" ]; then return; fi
@@ -315,23 +366,41 @@ classify_analyse_result_file ()
      while read line
      do
         # stage line
-	pat='^# .*(reset_artifacts|build_abe|build_llvm|benchmark)'
+	pat='^# .*(reset_artifacts|build_abe|build_llvm|benchmark|linux_n_obj)'
 	if [[ $line =~ $pat ]]; then
 	   stage="$(echo $line|sed -e 's|# ||' -e 's| --.*||' -e 's|:.*||')"
 	   #echo " $line => $stage"
 	fi
 
-        # Error line
+        # Clear error line
 	pat='^# Benchmarking infra is offline'
 	if [[ $line =~ $pat ]]; then
 	   set_diag "ERROR (infra offline)"
 	fi
-
-        # Error line
 	pat='# .* error: patch failed'
 	if [[ $line =~ $pat ]]; then
 	   set_diag "ERROR (git patch failed)"
 	fi
+	pat='# First few build errors in logs'
+	if [[ $line =~ $pat ]]; then
+	   set_diag "ERROR ($stage build errors)"
+	fi
+	pat='^# .*grew in size.*'
+	if [[ $line =~ $pat ]]; then
+	   set_diag "ERROR (grew in size)"
+	fi
+	pat='^# .*slowed down.*'
+	if [[ $line =~ $pat ]]; then
+	   set_diag "ERROR (slowed down)"
+	fi
+
+	# single message before reset-artifact
+	pat='^# FAILED'
+	if [[ $line =~ $pat ]] && [ ! -v stage ]; then
+	   set_diag "ERROR (FAILED in reset_artifacts)"
+	fi
+
+        [ ! -z "${test['diag']}" ] && break
 
      done < "${test['run_nb']}/artifact/artifacts/results"
 
@@ -396,12 +465,12 @@ print_suggestions ()
         echo "  --delete refs/heads/linaro-local/ci/$br \\"
       done
       echo ""
-    done 
+    done
     echo "rm -rf $tmpdir"
   fi
 }
 
-################## MAIN CLASSIFY FUNCTIONS 
+################## MAIN CLASSIFY FUNCTIONS
 classify_failures ()
 {
   local stale_jobs_file
@@ -420,8 +489,6 @@ classify_failures ()
 
   while read -r line;
   do
-     # For now, we classify only base-artifacts stale jobs
-     #[[ "$line" =~ ^base-artifacts: ]] || continue
 
      classify_get_project "$line"
 
@@ -474,10 +541,5 @@ if [ $classify != false ]; then
   exit 0;
 fi
 
+process_all_artifacts
 
-# Otherwise process the git to find not updated results
-for repo in "${repos[@]}"; do
-    process_git_url "$refs_url_prefix/$repo" &
-done | sort
-
-wait