Switch to relying on package names instead of module names #1976

OriolMunoz-da · 2025-08-20T13:19:26Z

Fixes #829

Pull Request Checklist

Cluster Testing

If a cluster test is required, comment /cluster_test on this PR to request it, and ping someone with access to the DA-internal system to approve it.
If a hard-migration test is required (from the latest release), comment /hdm_test on this PR to request it, and ping someone with access to the DA-internal system to approve it.

PR Guidelines

Include any change that might be observable by our partners or affect their deployment in the release notes.
Specify fixed issues with Fixes #n, and mention issues worked on using #n
Include a screenshot for frontend-related PRs - see README or use your favorite screenshot tool

Merge Guidelines

Make the git commit message look sensible when squash-merging on GitHub (most likely: just copy your PR description).

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da · 2025-08-21T14:46:49Z

/cluster_test

github-actions · 2025-08-21T14:47:15Z

Deploy cluster test triggered for Commit 3b845ce5b206cfe9ff5ed5af7ef95ee9acbc514e in , please contact a Contributor to approve it in CircleCI: https://app.circleci.com/pipelines/github/DACH-NY/canton-network-internal/28163

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da · 2025-08-25T15:09:37Z

apps/common/src/test/scala/org/lfdecentralizedtrust/splice/store/MultiDomainAcsStoreTest.scala

+      }
+    }
+
+    "prevent against ingestion of same (moduleName, entityName) with different package name - via interface fallback" in {


this test is the best summary for the reason of being of this PR

OriolMunoz-da · 2025-08-25T15:27:54Z

...main/resources/db/migration/canton-network/postgres/stable/V043__package_name_acs_stores.sql

+-- strictly not necessary, but lets us define the constraint as mandatory afterwards,
+-- and also makes index creation faster since there's no data to index
+truncate table acs_store_template cascade;
+truncate table user_wallet_acs_store cascade;
+truncate table external_party_wallet_acs_store cascade;
+truncate table validator_acs_store cascade;
+truncate table sv_acs_store cascade;
+truncate table dso_acs_store cascade;
+truncate table scan_acs_store cascade;
+truncate table splitwell_acs_store cascade;


double-check me that this is not a bad idea

note that all descriptors are bumped

OriolMunoz-da · 2025-08-25T15:28:26Z

apps/common/src/main/scala/org/lfdecentralizedtrust/splice/store/MultiDomainAcsStore.scala

-    * another list; that is why the contract ID by itself cannot be used by
-    * itself as the source of the sort key.
-    */
-  def listAssignedContractsNotOnDomainN(


unused except in a test (also deleted)

OriolMunoz-da · 2025-08-25T15:29:13Z

apps/common/src/main/scala/org/lfdecentralizedtrust/splice/store/db/AcsQueries.scala

      tableName: String,
      storeId: AcsStoreId,
      migrationId: Long,
-      where: SQLActionBuilder,
+      companion: C,


this should've been the case all along, instead of having where ti_q_n = ? on every query

agreed, thanks for pulling through that simplification!

OriolMunoz-da · 2025-08-25T15:29:53Z

apps/common/src/test/scala/org/lfdecentralizedtrust/splice/store/MultiDomainAcsStoreTest.scala

@@ -837,51 +837,6 @@ abstract class MultiDomainAcsStoreTest[
      MultiDomainAcsStoreTest.generatedCoids.value.size should (be >= 900 and be <= 1000)
    }

-    "read assignment-mismatched contracts in a stable order" in {


tested listAssignedContractsNotOnDomainN, deleted

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da · 2025-08-25T15:31:39Z

/cluster_test

github-actions · 2025-08-25T15:32:13Z

Deploy cluster test triggered for Commit 1680a64b4cb0ace131dd19435f78cb0d69064cea in , please contact a Contributor to approve it in CircleCI: https://app.circleci.com/pipelines/github/DACH-NY/canton-network-internal/28663

moritzkiefer-da

Very nice, thanks a lot!

Please run a performance test before merging so we can avoid surprises for long downtimes due to ACS backfilling.

moritzkiefer-da · 2025-08-26T06:06:38Z

apps/common/src/main/scala/org/lfdecentralizedtrust/splice/store/db/AcsQueries.scala

      tableName: String,
      storeId: AcsStoreId,
      migrationId: Long,
-      where: SQLActionBuilder,
+      companion: C,


agreed, thanks for pulling through that simplification!

moritzkiefer-da · 2025-08-26T06:13:10Z

...main/resources/db/migration/canton-network/postgres/stable/V043__package_name_acs_stores.sql

@@ -0,0 +1,362 @@
+-- strictly not necessary, but lets us define the constraint as mandatory afterwards,


Do we know what performance we get for a full ACS backfill? I'm a bit nervous about this given that the ACS queries are known to not perform well without pruning and there is no pruning atm on SV participants.

How about on CILR you take one SV (maybe best to use the SV runbook given that it resets daily anyway so if it goes wrong we can just reset again) manually force a reingestion by dropping the data from the table and last ingested offset?

If it does take too long, the easy short term option we have is to merge this into the 3.4 branch only. Afaict, there isn't an immediate need to have this be on 3.3.

How about on CILR you take one SV (maybe best to use the SV runbook given that it resets daily anyway so if it goes wrong we can just reset again) manually force a reingestion by dropping the data from the table and last ingested offset?

What is an "ACS backfill"? After truncating the ACS tables and bumping the store descriptor, the new ACS store will be initialized from the ACS endpoint of the ledger API, is that what you meant? Apart from the participant being slow to deliver the ACS, we are not batching SQL inserts when consuming the ACS

override def ingestAcs() = { ... // TODO (#989): batch inserts }

and we load the whole ACS into one in-memory structure:

private def ingestAcsAndInFlight(offset: Long) = { ... // TODO(#863): stream contracts instead of ingesting them as a single Seq

so I agree that one test on CILR is a good idea.

is that what you meant?

yes

rautenrieth-da · 2025-08-26T08:44:54Z

...main/resources/db/migration/canton-network/postgres/stable/V043__package_name_acs_stores.sql

+-- strictly not necessary, but lets us define the constraint as mandatory afterwards,
+-- and also makes index creation faster since there's no data to index
+truncate table store_last_ingested_offsets cascade;
+truncate table acs_store_template cascade;


This table should be empty, but it doesn't hurt to truncate it.

rautenrieth-da · 2025-08-26T08:49:35Z

...main/resources/db/migration/canton-network/postgres/stable/V043__package_name_acs_stores.sql

@@ -0,0 +1,362 @@
+-- strictly not necessary, but lets us define the constraint as mandatory afterwards,
+-- and also makes index creation faster since there's no data to index
+truncate table store_last_ingested_offsets cascade;


I don't think you can truncate this table, it's also used by txlog stores, see https://github.com/hyperledger-labs/splice/blob/main/apps/common/src/main/scala/org/lfdecentralizedtrust/splice/store/db/DbMultiDomainAcsStore.scala#L1080-L1086.

yep, you're right... I technically don't have to truncate it since I'm bumping the ACS descriptor, though then the "last ingested offset" for those would be incorrect
Since anyway I need to figure out how to safely* delete things on the SV runbook, I'll do the same here

*in the SV runbook, the descriptors don't change, so I can't leave last_ingested_offsets lingering after truncating

https://github.com/DACH-NY/canton-network-node/pull/18149

They will always be identical in practice, except when one of the stores is reset by changing the descriptor.

yeah that makes it hard for me to tell what I'm supposed to delete from the last_ingested 😢

actually nvm - I can just use the version numbers...

so, given this:

and this:

txLogStoreDescriptor = StoreDescriptor( version = 1, name = "DbScanStore", party = key.dsoParty, participant = participantId, key = Map( "dsoParty" -> key.dsoParty.toProtoPrimitive ), )

txLogStoreDescriptor = StoreDescriptor( version = 2, name = "DbUserWalletStore", party = key.endUserParty, participant = participantId, key = Map( "endUserParty" -> key.endUserParty.toProtoPrimitive, "validatorParty" -> key.validatorParty.toProtoPrimitive, "dsoParty" -> key.dsoParty.toProtoPrimitive, ), ),

I believe I can:

delete from store_last_ingested_offsets where store_id in (select id from store_descriptors where descriptor ->> 'name' IN -- these are ACS only, so we can delete everything ('DbSplitwellStore', 'DbSvSvStore', 'DbValidatorStore', 'DbExternalPartyWalletStore') -- these have txlog, preserve it. Version matches that of the txlog descriptor OR (descriptor ->> 'name' = 'DbUserWalletStore' AND descriptor ->> 'version' != 2) OR (descriptor ->> 'name' = 'DbScanStore' AND descriptor ->> 'version' != 1));

but I need an extra pair of eyes before i do anything stupid

Then again, that invalid package_name won't be used after bumping the descriptor (old data, unused, whenWeGetToIt ™️ deleted), so maybe that's not so bad?

how to handle package_name in that case

That is not a column in store_last_ingested_offsets? What I meant is you can truncate the acs tables, but not the store_last_ingested_offsets table. Then you can do whatever you want to the schema of the acs tables, and will only end up with one garbage row in store_descriptors and a few garbage rows in store_last_ingested_offsets (one per migration).

and will only end up with one garbage row in store_descriptors

it's not just garbage, but invalid, as the offset is wrong and the ACS won't be reingested if we rolled back to it - it's not clear to me that that's fine

Ah you're worried that the old data in the database would be corrupt, even if that data is not used? That's only a problem if we want to roll back the change to store descriptors relatively quickly, in order to use the old data before it becomes so outdated that it's useless? And since we're truncating the acs tables, there's no point doing a rollback, we'd have to restore from backup or do another store version bump, no?

copy pasting, this should be reasonable to do:

truncate all ACS_store tables, don't touch any other table

bump descriptors

indexes will be quick to create cause I just truncated

...allet/src/main/scala/org/lfdecentralizedtrust/splice/wallet/store/db/DbUserWalletStore.scala

Signed-off-by: Oriol Muñoz <[email protected]>

rautenrieth-da · 2025-08-27T08:14:22Z

...main/resources/db/migration/canton-network/postgres/stable/V043__package_name_acs_stores.sql

+-- where tablename like '%acs_store%'
+--   and indexdef like '%template_id_qualified_name%';
+DROP INDEX acs_store_template_sid_mid_tid_ce;
+CREATE INDEX acs_store_template_sid_mid_pn_tid_ce ON acs_store_template (store_id, migration_id, package_name,


This is a lot of indexes being created, did you time it on CILR? If it's more than a minute we should probably mention it in the release notes. There's also SqlIndexInitializationTrigger for creating expensive indexes asynchronously, but having the apps start without fast ACS queries is probably asking for trouble.

took less than a minute to create all indexes for a DSO store copy, which is the biggest one according to https://grafana.cilr.global.canton.network.digitalasset.com/d/dduss3xr5or28c/splice-store-acs-size?orgId=1&from=now-1h&to=now&timezone=UTC&var-namespace=sv-1&var-store_name=$__all
(total of 300k rows across all store ids)

a lot less, if I end up managing to truncate the tables

rautenrieth-da · 2025-08-27T08:16:18Z

apps/common/src/main/scala/org/lfdecentralizedtrust/splice/store/db/AcsQueries.scala

@@ -326,12 +356,22 @@ object AcsQueries {
        companionClass: ContractCompanion[C, TCId, T],
        decoder: TemplateJsonDecoder,
    ): Contract[TCId, T] = {
+      // safety check: if the templates don't match,


Suggested change

// safety check: if the templates don't match,

// safety check: if the packages don't match,

?

// safety check: if the PackageQualifiedNames don't match,

is most correct, will fix

OriolMunoz-da · 2025-08-27T13:56:07Z

timings for ACS reingestion:

a bit more than 2m for scan, logs:

and SV app doesn't reingest, see #2041

OriolMunoz-da added 9 commits August 20, 2025 12:19

get a failing test

ff086b9

Signed-off-by: Oriol Muñoz <[email protected]>

Break when the results would be bad [ci]

ea2f40a

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] fix things

ee72588

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] more fixes

cae3fad

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] was that a gha/k8s failure or did i cause that? let's find out

4199050

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] refactor so many queries something is bound to break

0c01122

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] remove many instances of getFromResources

096ea57

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] fix splitwell

24b712a

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] fix bad query

3b845ce

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da added 6 commits August 21, 2025 15:40

unused variable

50077da

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] bump descriptors

15d666a

Signed-off-by: Oriol Muñoz <[email protected]>

Merge branch 'main' into oriol/packagename-stores-interface-fallback

a76111d

fix many compile errors

59c5890

Signed-off-by: Oriol Muñoz <[email protected]>

add indexes with todo

3184d5a

Signed-off-by: Oriol Muñoz <[email protected]>

[ci] change definition of all indexes involving tid

5910a9f

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da changed the title ~~Oriol/packagename stores interface fallback~~ Switch to relying on package names instead of module names Aug 25, 2025

OriolMunoz-da commented Aug 25, 2025

View reviewed changes

[ci] also truncate store_last_ingested_offsets

1680a64

Signed-off-by: Oriol Muñoz <[email protected]>

OriolMunoz-da marked this pull request as ready for review August 25, 2025 15:31

OriolMunoz-da requested review from rautenrieth-da and moritzkiefer-da August 25, 2025 15:31

moritzkiefer-da approved these changes Aug 26, 2025

View reviewed changes

rautenrieth-da reviewed Aug 26, 2025

View reviewed changes

Merge branch 'main' into oriol/packagename-stores-interface-fallback

dc7e00d

OriolMunoz-da commented Aug 26, 2025

View reviewed changes

...allet/src/main/scala/org/lfdecentralizedtrust/splice/wallet/store/db/DbUserWalletStore.scala Outdated Show resolved Hide resolved

[ci] revert bump of txlog descriptor

5bfdbca

Signed-off-by: Oriol Muñoz <[email protected]>

rautenrieth-da reviewed Aug 27, 2025

View reviewed changes

OriolMunoz-da mentioned this pull request Aug 27, 2025

Handle bumping acs store descriptors to already existing versions #2033

Open

		@@ -0,0 +1,362 @@
		-- strictly not necessary, but lets us define the constraint as mandatory afterwards,

	// safety check: if the templates don't match,
	// safety check: if the packages don't match,

Switch to relying on package names instead of module names #1976

Are you sure you want to change the base?

Switch to relying on package names instead of module names #1976

Uh oh!

Conversation

OriolMunoz-da commented Aug 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Pull Request Checklist

Cluster Testing

PR Guidelines

Merge Guidelines

Uh oh!

OriolMunoz-da commented Aug 21, 2025

Uh oh!

github-actions bot commented Aug 21, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

OriolMunoz-da commented Aug 25, 2025

Uh oh!

github-actions bot commented Aug 25, 2025

Uh oh!

moritzkiefer-da left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

OriolMunoz-da Aug 26, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

OriolMunoz-da Aug 27, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

OriolMunoz-da commented Aug 20, 2025 •

edited

Loading

OriolMunoz-da Aug 26, 2025 •

edited

Loading

OriolMunoz-da Aug 27, 2025 •

edited

Loading