Hide storing ephemeral nodes behind a per-tree flag

CHANGELOG.md

@@ -2,12 +2,7 @@
 
 ## HEAD
 
-### Storage
-
-* #2568: Ephemeral nodes of the Merkle tree are no longer written to storage.
-  This is backwards compatible with versions >= v1.4.0. The leftover ephemeral
-  nodes may continue being stored, but since they are not used by reads, the API
-  remains operating correctly.
+* #2568: Allow disabling the writes of ephemeral nodes to storage.
 
 ## v1.4.1
 

integration/format/testdata/dump_tree_output_1000

@@ -12,11 +12,23 @@ leaves: {
   key: "CAI="
   value: "\xbb\x91zJ\xbb\xbbR\xc7\xe0>\xde\x1e\xe5\x9fC\x11\xe5J\x85\x0b\x00\x99\xd1^\xa2\xd8RB?;S5"
 }
+leaves: {
+  key: "CAM="
+  value: "\xc3\xe2VW\xbb\xed\x8f\x1eۂz\x05\xf1\x86;\xc21\x05\xae\x05\x8e\x05-o9\xb5\x05*|\x9b\x1c@"
+}
+internal_nodes: {
+  key: "BgA="
+  value: "7ה\x03g\xb8+)\xe4\n\r\xb6\x9a\x83\xf8\x11}ui\xe7\xf7N\xc1:\x87S\x06%\\'\x7f\x1b"
+}
 internal_nodes: {
   key: "BwA="
   value: "\xc6\xee@\x91R\xb3TI\x03\xf2\x86B-}c:\x9aۢ\xef\x0f/\xdb\xf1V\x8e-\r\xcf\x17?\x92"
 }
-internal_node_count: 1
+internal_nodes: {
+  key: "BwI="
+  value: "\xe3\xda%\xfem\x86\x03v#T\x8e0\xf8 \x1e\xdb\xfb\xad\xef\\\x18\x8asX\xec\x0fN\xbe\"\x901\x1e"
+}
+internal_node_count: 3
 
 prefix: "\x00\x00\x00\x00\x00\x00\x00"
 depth: 8
@@ -7104,6 +7116,10 @@ internal_nodes: {
   key: "AQA="
   value: "\xb2\nE\x80\x9e\x88X\xb4\xe5\x14\xa7ic\x80\xd1b**.u\n\xfd \t\x90e\xae\xc4\xee\x89Q\xdc"
 }
+internal_nodes: {
+  key: "AYA="
+  value: "\x8e\xcc\xe2\x04\x1c}\x0b\x0f\x1f\xa0K\xc4y\xb3mӠP\x8d\xa68\x02\x178\xf2@\xac\xd3\xd1/&\x8f"
+}
 internal_nodes: {
   key: "AgA="
   value: "A:\x85Z\x8aıqBκ\t9\x02\x0fF\xe9\x91s\x84\x07\"\x08\x08Ů\xfe\xea-\xf6\xe8R"
@@ -7116,6 +7132,10 @@ internal_nodes: {
   key: "AoA="
   value: "\xb1P\x17A\x02\xe2\xe40\xff\x16q\x9e\x1d\x91\xf6\xf0\x0e\x81\xf6\x9a\x81\xd0\r/\xac\x0e4]?3\x12r"
 }
+internal_nodes: {
+  key: "AsA="
+  value: "\xf0ݟo9d\x0f\x8f𠅼\r\x85\x07k\rV2\x84\x99 \xeb\x0c4\x06\xdd\xc4\xf7K\xf2\x98"
+}
 internal_nodes: {
   key: "AwA="
   value: "\xf9ϧ\x1d\xf4?Zٶ\x82\xfc\\g\xc9\xf2eyĭ\x03\x0eI\rX\xc5\xdd\xd4\xf7\\K\xd0G"
@@ -7992,5 +8012,5 @@ internal_nodes: {
   key: "Bzw="
   value: "\xd5u\xac\xe5b\xa0&\xaa\x92\xd9\xf6~FO\x9c\xa2>c\x80\xe3>rI\x86\xe5\x91'l\xaf"
 }
-internal_node_count: 228
+internal_node_count: 230
 

integration/format/testdata/dump_tree_output_871

@@ -12,11 +12,19 @@ leaves: {
   key: "CAI="
   value: "\xbb\x91zJ\xbb\xbbR\xc7\xe0>\xde\x1e\xe5\x9fC\x11\xe5J\x85\x0b\x00\x99\xd1^\xa2\xd8RB?;S5"
 }
+internal_nodes: {
+  key: "BgA="
+  value: "\xf6\x87\xc1\xbdh\x94\xb4Q\xdc\xf7\x1fZ+\x03\x9d\xce`\xf2\x87Џ\x89\x02\xc1iw\xa3AC\xbe\xa7N"
+}
 internal_nodes: {
   key: "BwA="
   value: "\xc6\xee@\x91R\xb3TI\x03\xf2\x86B-}c:\x9aۢ\xef\x0f/\xdb\xf1V\x8e-\r\xcf\x17?\x92"
 }
-internal_node_count: 1
+internal_nodes: {
+  key: "BwI="
+  value: "\x07U\xae&\xaf\xea\xbd1\x10܍\xe6\xf0\xac\xf1\xe5)\x02\xddk\x80Q`/P\\\xa9\xe7\xfa_\xc3/"
+}
+internal_node_count: 3
 
 prefix: "\x00\x00\x00\x00\x00\x00\x00"
 depth: 8
@@ -6568,10 +6576,18 @@ internal_nodes: {
   key: "A0A="
   value: "K\x1a\xbc\x93\xb8[20.Җ`\x9d\x12\x1f\x90W\xa3Ȥ~\r>\t0\x02\xfd\x07\tL\x03y"
 }
+internal_nodes: {
+  key: "AQA="
+  value: "$F\xb2T\xeb\x8c\x03\xf2\xef`3ۄ\xbfik\xba\x97\xb5\x89\x1c^\x81\xfa4\x0e\xf8\xf5#\"ӑ"
+}
 internal_nodes: {
   key: "AgA="
   value: "A:\x85Z\x8aıqBκ\t9\x02\x0fF\xe9\x91s\x84\x07\"\x08\x08Ů\xfe\xea-\xf6\xe8R"
 }
+internal_nodes: {
+  key: "AkA="
+  value: "\xf5<\x8c8C\n\x1c\xa6\xdbG\x94\xbaL\x85\xac\xf0x\xeb\xb2\x03i\xaf}\xeb\x1b\xa8\x96\x99\x19z>\xab"
+}
 internal_nodes: {
   key: "AwA="
   value: "\xf9ϧ\x1d\xf4?Zٶ\x82\xfc\\g\xc9\xf2eyĭ\x03\x0eI\rX\xc5\xdd\xd4\xf7\\K\xd0G"
@@ -6728,6 +6744,10 @@ internal_nodes: {
   key: "BVg="
   value: "5W\x98P\xfe\x125'\xb9K0\xcf\x08\x17\x92\xdb\xfb\x04\xda^\xf6\x03K\x04B\xd9ifHZ\xb6\xfe"
 }
+internal_nodes: {
+  key: "BWA="
+  value: "\xd6N^\x93e\x13Qd\xe1\xea\xc0_a\x04\x8dL\x9fu\xad\x05\xe6\xd1y&\xa3\xa1\xc3Et\xfd\xa7\xde"
+}
 internal_nodes: {
   key: "BgA="
   value: "\xec\x86&v\xd5C*\x15\xb3\xec\t\x89_\xf1*\xc1\xf02Ұ\xbdk\x08\xb0x\xfc\xdfȥ\xeaܾ"
@@ -6828,6 +6848,10 @@ internal_nodes: {
   key: "BmA="
   value: "\x15\xd6l\x92\xd7$*q\x9f\xbf\xd6\x7f\x88\xb0\xc2`\x1d\xc2y\xc3\xfe\x82bԥ\xc4\xe9\r}i\x97!"
 }
+internal_nodes: {
+  key: "BmQ="
+  value: "d\xbct\x8f\xdema\x7f\xd3\xf1\\\x14/l\xb3{{\x02\x87\xc4>\xdd,+\xfe\x87\x7f\xc8PTJ\xa7"
+}
 internal_nodes: {
   key: "Bw4="
   value: "\x99\xbc\xce}\xcd\xf2\xfe\xb9]\xecY:P7\x05~l\xfc\xab\x066 G\xa1y\x9c\x9a\xe4\xe1\r\x89\xb9"
@@ -6956,5 +6980,5 @@ internal_nodes: {
   key: "Bzw="
   value: "\xd5u\xac\xe5b\xa0&\xaa\x92\xd9\xf6~FO\x9c\xa2>c\x80\xe3>rI\x86\xe5\x91'l\xaf"
 }
-internal_node_count: 98
+internal_node_count: 102
 

integration/format/testdata/dump_tree_output_96

@@ -388,6 +388,10 @@ internal_nodes: {
   key: "A0A="
   value: "\xa2\x9b|=$\xce/\x86+\x0b\"\x92n\xcc\xc3\\/b7Lm\xdb\xd5\x19{\xfeKo\xe8\xbdtF"
 }
+internal_nodes: {
+  key: "AQA="
+  value: "\xf0K\xff\xed19\x0e\tL\xea\xfc\x12U\xf1\xf2α1z\x8a\xba\xe0\xb0t\x9a\x88\x15]U\xee\xce\xd0"
+}
 internal_nodes: {
   key: "AgA="
   value: "\x8f\x05[\xe4\x9bG\x8fW\x12\xeb\xe2\xd8[\x89˔m\xccP\x87\x18_\x1d.\xb66S\xa5\x99\x81\x08\xce"
@@ -760,5 +764,5 @@ internal_nodes: {
   key: "Bzw="
   value: "\xd7\x1a\xff\x1a\xea\xde\xc2\xe8\xe8<p;9\x16\x1f\x10\xde\xd4?\x0e\xe6\xdcʗ+\x8b\x0cf >7d"
 }
-internal_node_count: 94
+internal_node_count: 95
 

log/sequencer.go

@@ -17,8 +17,10 @@ package log
 import (
 	"bytes"
 	"context"
+	"flag"
 	"fmt"
 	"strconv"
+	"strings"
 	"sync"
 	"time"
 
@@ -64,6 +66,22 @@ var (
 	QuotaIncreaseFactor = 1.1
 )
 
+// The tree IDs for which sequencer does not store ephemeral node hashes.
+// Trillian releases up to v1.4.1 store the ephemeral hashes, which corresponds
+// to this slice being empty. Release v1.4.2 allows disabling this behaviour
+// for individual, or all trees (denoted by the "*" wildcard). The release
+// after v1.4.2 will switch to "*" behaviour unconditionally.
+var idsWithNoEphemeralNodes = make(map[string]bool)
+
+// TODO(pavelkalinnikov): Remove this flag in the next release.
+func init() {
+	var ids string
+	flag.StringVar(&ids, "tree_ids_with_no_ephemeral_nodes", "", "Comma-separated list of tree IDs for which storing the ephemeral nodes is disabled, or * to disable it for all trees")
+	for _, id := range strings.Split(ids, ",") {
+		idsWithNoEphemeralNodes[id] = true
+	}
+}
+
 func quotaIncreaseFactor() float64 {
 	if QuotaIncreaseFactor < 1 {
 		QuotaIncreaseFactor = 1
@@ -185,8 +203,18 @@ func updateCompactRange(cr *compact.Range, leaves []*trillian.LogLeaf, label str
 			return nil, nil, err
 		}
 	}
-	// Note: The ephemeral node hashes are not stored.
-	hash, err := cr.GetRootHash(nil)
+
+	// Store or not store ephemeral nodes depending on the flag. This is a
+	// temporary safety measure, to test this on individual trees before fully
+	// disabling ephemeral nodes.
+	storeEphemeral := store
+	if idsWithNoEphemeralNodes[label] || idsWithNoEphemeralNodes["*"] {
+		storeEphemeral = nil
+	}
+
+	// TODO(pavelkalinnikov): Do not store the ephemeral node hashes
+	// unconditionally, because they are not used since v1.4.0.
+	hash, err := cr.GetRootHash(storeEphemeral)
 	if err != nil {
 		return nil, nil, err
 	}

log/sequencer_test.go

@@ -154,6 +154,11 @@ var (
 	})
 )
 
+func init() {
+	// The tree ID used by TestIntegrateBatch.
+	idsWithNoEphemeralNodes["154035"] = true
+}
+
 func makeSLR(root *types.LogRootV1) *trillian.SignedLogRoot {
 	logRoot, _ := root.MarshalBinary()
 	return &trillian.SignedLogRoot{LogRoot: logRoot}