diff --git a/BACKLOG.md b/BACKLOG.md
index 2c78d3a..f8d5efe 100644
--- a/BACKLOG.md
+++ b/BACKLOG.md
@@ -2027,7 +2027,7 @@ lent, partition`); the leader tells each named node's agent (root) what to be, t
 - Phases: P1 the `slurm` package (conf, key, parsers) and settings; P2 the agent side (proto, install, munge,
   slurmd/slurmctld, status); P3 the leader side (conf from reports, lend/reclaim); P4 `ctl slurm`
   (status, lend, reclaim) and a TUI screen; P5 `ctl train import` of a Slurm job's output into the mirror.
-  Later: a backup slurmctld on a second node, accounting (slurmdbd on the Patroni PG), users from LDAP/AD.
+  Later: a backup slurmctld on a second node, accounting (slurmdbd; it takes MySQL/MariaDB only, not the Patroni PG), users from LDAP/AD.
 - Tried on the PoC 2026-11-11 (dev build v0.16.3-17-g849760c-slurm on 140, 150, node05, node06; Ubuntu
   24.04, Slurm 23.11.4 from the archive): `ctl slurm setup --controller llmgw-node06 --nodes llmgw-node05`
   installed munge, slurmctld and slurmd, node05 entered drained with its slurmd -C hardware (8 CPUs,
@@ -2037,7 +2037,9 @@ lent, partition`); the leader tells each named node's agent (root) what to be, t
   from the daemon); the packaged unit is Type=notify (slurmd needs --systemd); slurmd's own "slurmstepd
   infinity" was counted as a job; without accounting storage 23.11's main scheduler marks batch jobs
   InvalidAccount and backfill starts them, one a pass (bf_interval=10: 6-10 s each). So many small jobs
-  want slurmdbd. Not tried yet: a GPU node (gres).
+  want slurmdbd. — Wrong cause, found 2026-11-11: a job asking no memory got the whole node's (ReqTRES
+  mem=all), so a node ran one job at a time and each next one waited for backfill's pass. Fixed: the
+  partition's DefMemPerCPU is the leanest node's share; slurmdbd is not needed for this. Not tried yet: a GPU node (gres).
 - Failover while lent, tried 2026-11-11 (v0.16.3-23): the first try killed the job (NODE_FAIL). A new
   leader had heard from no node yet and sent the controller a slurm.conf without them; slurmctld,
   restarted on it, dropped node05 and its job. Fixed (141a1f4, 0774e90): slurm.conf's nodes and the
diff --git a/internal/slurm/slurm.go b/internal/slurm/slurm.go
index 594a0d7..741e159 100644
--- a/internal/slurm/slurm.go
+++ b/internal/slurm/slurm.go
@@ -219,8 +219,7 @@ GresTypes=gpu
 MpiDefault=none
 # No mail sent; without it every start says MailProg is invalid.
 MailProg=/bin/true
-# Without accounting storage, 23.11's main scheduler marks a batch job
-# InvalidAccount; backfill starts it on its next pass, so it is kept short.
+# Backfill looks again every 10 s, not every 30.
 SchedulerParameters=bf_interval=10
 # A node that went down and comes back is taken again; one llmgw drained
 # stays drained until llmgw resumes it.
@@ -230,6 +229,15 @@ SlurmdTimeout=300
 `)
 	nodes := c.ready()
 	names := make([]string, 0, len(nodes))
+	// A job that asks for no memory gets this much per CPU: the share of
+	// the leanest node. Without it it gets the whole node's, and the jobs
+	// on a node run one at a time.
+	var perCPU int64
+	for _, n := range nodes {
+		if share := int64(float64(n.RealMemory)*memoryKept) / int64(n.CPUs); perCPU == 0 || share < perCPU {
+			perCPU = share
+		}
+	}
 	for _, n := range nodes {
 		names = append(names, n.Name)
 		fmt.Fprintf(&b, "NodeName=%s NodeAddr=%s CPUs=%d", n.Name, n.Addr, n.CPUs)
@@ -250,8 +258,8 @@ SlurmdTimeout=300
 		fmt.Fprintf(&b, " State=DRAIN Reason=%q\n", ReasonNew)
 	}
 	if len(names) > 0 {
-		fmt.Fprintf(&b, "PartitionName=%s Nodes=%s Default=YES MaxTime=INFINITE State=UP\n",
-			c.Partition, strings.Join(names, ","))
+		fmt.Fprintf(&b, "PartitionName=%s Nodes=%s Default=YES MaxTime=INFINITE State=UP DefMemPerCPU=%d\n",
+			c.Partition, strings.Join(names, ","), perCPU)
 	}
 	return b.String()
 }
diff --git a/internal/slurm/slurm_test.go b/internal/slurm/slurm_test.go
index 4511b91..033e525 100644
--- a/internal/slurm/slurm_test.go
+++ b/internal/slurm/slurm_test.go
@@ -82,6 +82,8 @@ func TestSlurmConf(t *testing.T) {
 		"NodeName=cuda01 NodeAddr=10.0.0.147 CPUs=16 RealMemory=62080 Gres=gpu:rtx3090:1 State=DRAIN Reason=\"llmgw: not lent yet\"\n",
 		"NodeName=site2 NodeAddr=10.1.0.5 CPUs=32 Boards=1 SocketsPerBoard=2 CoresPerSocket=8 ThreadsPerCore=2 RealMemory=97000 Gres=gpu:t4:1,gpu:v100:2 State=DRAIN Reason=\"llmgw: not lent yet\"\n",
 		"PartitionName=gpu Nodes=cuda01,site2 Default=YES",
+		// site2's share, the leaner: 97000 MiB for 32 CPUs.
+		" DefMemPerCPU=3031\n",
 	} {
 		if !strings.Contains(conf, want) {
 			t.Errorf("slurm.conf has no %q:\n%s", want, conf)