Backups: a revision chain belongs to a PC, not just a machine

plugins/backups/plugin.py

@@ -25,7 +25,7 @@ from .api import backups_bp
 from .models import BackupRevision
 from .services.registry import (REGISTRY, getkind, canonicalhash,
                                DEFAULTSHAREROOT)
-from .services.retention import prune
+from .services.retention import prune, samesource as _samesource
 
 logger = logging.getLogger(__name__)
 
@@ -266,15 +266,26 @@ class BackupsPlugin(BasePlugin):
                     "contenthash and sharepath are required for kind '{}'".format(
                         kindkey))
 
-        # with_for_update serializes concurrent posts for the same asset+kind.
-        # The whole fleet collects on the same GE-Enforce cycle, so a retry or
-        # two PCs claiming one machine number can otherwise both pass the hash
-        # check and insert duplicate identical revisions. No unique constraint
-        # can cover this: dedup is against the LATEST row only, because a config
+        # Dedup is against the LATEST row of THIS SOURCE's chain, not the
+        # asset's. Several PCs legitimately share one machine number here - the
+        # part markers on 0613, 0615 and WJPRT do, and their configs genuinely
+        # differ, typically by COM port. Keyed on
+        # asset alone, each marker's post differed from whichever marker posted
+        # last, so nothing ever deduped and the chain grew by one revision per
+        # PC per collection cycle. Keyed on the source, an unchanged config is a
+        # no-op again and each marker keeps its own history against the machine.
+        #
+        # Dedup is against the latest row only, not any row, because a config
         # reverting to an earlier state is a legitimate new revision.
+        #
+        # with_for_update serializes concurrent posts for the same chain: the
+        # fleet collects on one GE-Enforce cycle, so a retry could otherwise
+        # pass the hash check twice and insert the same revision.
+        sourcehostname = payload.get('sourcehostname')
         latest = (db.session.query(BackupRevision)
                   .filter(BackupRevision.assetid == assetid,
-                          BackupRevision.backupkind == kindkey)
+                          BackupRevision.backupkind == kindkey,
+                          _samesource(sourcehostname))
                   .order_by(BackupRevision.backuprevisionid.desc())
                   .with_for_update()
                   .first())