From dd110efb1eb7ff9362f5eeb5d6d71243267e8521 Mon Sep 17 00:00:00 2001
From: Matyas Selmeci <matyas@cs.wisc.edu>
Date: Wed, 18 Dec 2024 18:10:23 -0600
Subject: [PATCH 1/3] Check if a mirror has migrated to the distrepos format
 before adding it to the mirrorlist

---
 distrepos/mirror_run.py | 32 ++++++++++++++++++++++++++++++--
 1 file changed, 30 insertions(+), 2 deletions(-)

diff --git a/distrepos/mirror_run.py b/distrepos/mirror_run.py
index 8d28e76..47fbd40 100644
--- a/distrepos/mirror_run.py
+++ b/distrepos/mirror_run.py
@@ -45,6 +45,30 @@ def get_mirror_info_for_arch(hostname: str, tag: Tag, arch: str) -> t.Tuple[str,
     return mirror_base, repomd_url
 
 
+def is_migrated(repomd_url: str) -> bool:
+    """
+    Given the URL of a repomd file, check if the repo that owns that file
+    has been migrated to the distrepos format.  This is true if they have
+    a file named "pkglist" as a sibling of the "repodata" folder (which is
+    the parent of the "repomd.xml" file.
+
+    Args:
+        repomd_url: The URL of a repomd.xml file
+
+    Returns: True if the repo is in the distrepos format.
+    """
+    pkglist_url = os.path.dirname(os.path.dirname(repomd_url)) + "/pkglist"
+    _log.info(f"Checking for mirror format based on {pkglist_url}")
+    response = requests.get(pkglist_url, timeout=10)
+    if response.status_code == 404:
+        _log.info(f"pkglist file not found; repo probably not migrated")
+        return False
+    elif response.status_code != 200:
+        _log.warning(f"unexpected response.code for pkglist file {pkglist_url}: {response.status_code}")
+        return False
+    return True
+
+
 def test_single_mirror(repodata_url: str) -> bool:
     """
     Given the full URL of a repodata/repomd.xml that might mirror a tag, return whether
@@ -75,8 +99,12 @@ def test_single_mirror(repodata_url: str) -> bool:
             )
             return False
         else:
-            _log.debug(f"Mirror {repodata_url} all good")
-            return True
+            if is_migrated(repodata_url):
+                _log.debug(f"Mirror {repodata_url} all good")
+                return True
+            else:
+                _log.debug(f"Mirror {repodata_url} not migrated")
+                return False
 
 
 def update_mirrors_for_tag(options: Options, tag: Tag) -> t.Tuple[bool, str]:

From c5eb7adfced318cd194b437edef2a3b76823cefc Mon Sep 17 00:00:00 2001
From: Matyas Selmeci <matyas@cs.wisc.edu>
Date: Fri, 27 Dec 2024 10:31:38 -0600
Subject: [PATCH 2/3] Add final newline to mirrorlist files

---
 distrepos/mirror_run.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/distrepos/mirror_run.py b/distrepos/mirror_run.py
index 47fbd40..02960eb 100644
--- a/distrepos/mirror_run.py
+++ b/distrepos/mirror_run.py
@@ -144,7 +144,7 @@ def update_mirrors_for_tag(options: Options, tag: Tag) -> t.Tuple[bool, str]:
         working_path.mkdir(parents=True, exist_ok=True)
 
         with open(working_path / arch, 'w') as mirrorf:
-            mirrorf.write('\n'.join(good_mirrors))
+            mirrorf.write('\n'.join(good_mirrors) + '\n')
 
     update_release_repos(dest_path, working_path, prev_path)
 

From 7ca0e49061557f1999a5aed1e50f09fdbf75a9d1 Mon Sep 17 00:00:00 2001
From: Matyas Selmeci <matyas@cs.wisc.edu>
Date: Fri, 27 Dec 2024 10:35:01 -0600
Subject: [PATCH 3/3] Uncomment mirror hosts -- it's OK to have them in the
 list since we check for whether they migrated

---
 etc/distrepos.conf | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/etc/distrepos.conf b/etc/distrepos.conf
index 97a7b6c..c73b1b8 100644
--- a/etc/distrepos.conf
+++ b/etc/distrepos.conf
@@ -94,13 +94,11 @@ static_root = /data/repo/osg/archive/repo
 mirror_root = /data/mirror
 
 # A list of known mirror hosts
-# XXX Comment mirror hosts back in, individually, once their admins have
-#     reported migrating to the new layout.
-#mirror_hosts =
-#    https://t2.unl.edu
-#    https://mirror.grid.uchicago.edu/pub
-#    https://linux-mirrors.fnal.gov/linux/
-#    http://mirror.hep.wisc.edu/upstream
+mirror_hosts =
+    https://t2.unl.edu
+    https://mirror.grid.uchicago.edu/pub
+    https://linux-mirrors.fnal.gov/linux/
+    http://mirror.hep.wisc.edu/upstream
 
 
 # Parent directory within dest_root for tarball client