Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix(chainindex): retry transaction if database connection is lost #12657

Merged
merged 3 commits into from
Oct 29, 2024

Conversation

aarshkshah1992
Copy link
Contributor

Retry transaction if database connection is lost.

@aarshkshah1992 aarshkshah1992 added the skip/changelog This change does not require CHANGELOG.md update label Oct 29, 2024
@@ -144,7 +144,7 @@ func withTx(ctx context.Context, db *sql.DB, fn func(*sql.Tx) error) error {
}

func isRetryableError(err error) bool {
return err != nil && strings.Contains(err.Error(), "database is locked")
return err != nil && (strings.Contains(err.Error(), "database is locked") || strings.Contains(err.Error(), "statement is closed"))
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not convinced this is the right approach, why is the statement getting closed? it's not a network connection so shouldn't suffer from network problems. This seems like it's masking a real problem.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@rvagg I see what you mean. Have reverted this change.

@aarshkshah1992
Copy link
Contributor Author

@rvagg Do you mean to say that there should never be a transient conn error with Sqllite ?

@rvagg
Copy link
Member

rvagg commented Oct 29, 2024

I can't imagine why. Unless the docs suggest otherwise? It's basically an open file descriptor?

@aarshkshah1992
Copy link
Contributor Author

@rvagg I see what you mean. I have reverted that change.

I have a strong suspicion based on meandering through the code that the RPC context cancellation from the client side (say a transient RPC connection failure) while a statement is executing is causing this.

I've added some logging to look this up the next time we run into this.

@aarshkshah1992 aarshkshah1992 merged commit 5489ae9 into feat/msg-eth-tx-index Oct 29, 2024
74 of 75 checks passed
@aarshkshah1992 aarshkshah1992 deleted the feat/retry-db-conn-lost branch October 29, 2024 14:53
aarshkshah1992 added a commit that referenced this pull request Oct 31, 2024
* chain index complete for msgs and txns

* dont need observer changes for now

* changes

* fix tests

* fix tests

* use th right context

* index empty tipsets correctly

* implement automated backfilling

* add event indexing and remove all old indices

* fix test

* revert deployment test changes

* revert test changes and better error handling for eth tx index lookups

* fix sql statments naming convention

* address review for Index GC

* more changes as per review

* changes as per review

* fix config

* mark events as reverted during reconciliation

* better reconciliation; pens down and code complete; also reconcile events

* fix tests

* improve config and docs

* improve docs and error handling

* improve read logic

* improve docs

* better logging and handle ennable event storage

* improve logs and index init proc

* better logging

* fix bugs based on calibnet testing

* create sqliite Indices

* gc should be based on epochs

* fix event query

* foreign keys should be enabled on the DB

* reverted tipsets should be removed as part of GC

* release read lock

* make it easy to backfill an empty index using reconciliation

* better docs for reconciliation

* fix conflicts with master

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix go mod

* fix formatting

* revert config changes

* address changes in observer

* remove top level chainindex package

* changes as per review

* changes as per review

* changes as per review

* handle index with reverted tipsets during reconciliation

* changes as per review

* fix type of max reconcile epoch

* changes to reconciliation as per review

* log ipld error

* better logging of progress

* disable chain indexer hydrate from snapshot based on config

* always populate index

* make config easy to reason about

* fix config

* fix messaging

* revert config changes

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* changes as per review

* make error messages homogenous

* fix indentation

* changes as per review

* feat: recompute tipset to generate missing events if event indexing is enabled (#12463)

* auto repair events

* make jen

* fix leaky abstraction

* better docs for gc retention epoch

* imrpove DB handling (#12485)

* fix conflict

* fix lite node config for indexer

* exclude reverted events from eth get logs if client queries by epoch

* Simply addressing for event lookups in the index.

simply addressing for event lookups

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix tests

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* feat: migration("re-indexing"), backfilling and diasgnostics tooling for the `ChainIndexer` (#12450)

* fix conflicts with chain indexer

* feat: chain indexer todos [skip changelog] (#12462)

* feat: finish todos of validation api

* feat: add indexed data verification with chain store

* feat: address comments and finish TODO

* fix: build issue

* address comments

* fix: ci issue

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* changes to Index Validation API based on Rodds first review

* build chain indexer API

* improve error handling

* feat: lotus-shed tooling for chain indexer (#12474)

* feat: add lotus-shed command for backfilling chain indexer

* feat: add lotus-shed command for inspecting the chain indexer

* feat: use single lotus-shed command to inspect and backfill

* fix: remove the unused queries

* small changes

* add change log

* backfilling improvements and fixes

* finish chain index validation and backfill tooling

* user documentation for the

* validate from epoch

* Apply suggestions from code review

Suggestions from Steve's read of the user doc.

Co-authored-by: Steve Loeppky <[email protected]>

* changes to user doc as per review

* Apply suggestions from code review

Co-authored-by: Steve Loeppky <[email protected]>

* changes to user doc as per review

* Apply suggestions from code review

Co-authored-by: Steve Loeppky <[email protected]>

* changes as per review

* feat: add event entries count in validation API (#12506)

* feat: add event entry count in validation API

* address comments

* use sqllite defaults (#12504)

* Apply suggestions from code review

Co-authored-by: Steve Loeppky <[email protected]>

* write chain index to a different dir

* Apply suggestions from code review

Co-authored-by: Steve Loeppky <[email protected]>

* fix conflicts

* UX improvements to backfilling

* feat: tests for the chain indexer (#12521)

* ddl tests

* tests for the chain indexer

* finish unit tests for chain indexer

* fix formatting

* cleanup reverted tipsets to avoid db bloat

* fix logging

* test for filter by address

* test gc cascade delete

* fix db locked error during backfilling

* fix var name

* increase db locked timeout

* fix db locked issue

* reduce db lock timeout

* no lock in gc

* reconcile does not need lock

* improved error handling

* Update chain-indexing-overview-for-rpc-providers.md

Doc updates based on @jennijuju feedack.

* Update chain-indexing-overview-for-rpc-providers.MD

Fixes after reviewing 33c1ca1

* better metrics for backfilling

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* Update chain/index/chain-indexing-overview-for-rpc-providers.MD

Co-authored-by: Rod Vagg <[email protected]>

* tests for changes to event addressing

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* changes as per review -> round 1

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* log tipset key cid

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix docs

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix tests

* fix tests

* make jen

* fix conflicts

---------

Co-authored-by: Aryan Tikarya <[email protected]>
Co-authored-by: Rod Vagg <[email protected]>
Co-authored-by: Steve Loeppky <[email protected]>

* fix lint

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* remove reverted flag from RPC

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix testing of events and dummy chain store

* remove lotus shed commands for old Indices

* change type of event counts to uint64

* only recompute events if theyre not found

* short-circuit empty events path for older tipsets

* chain indexer must be enabled if ETH RPC is enabled

* change name of message_id column to id in tipset_message table

* only expose SetRecomputeTipSetStateFunc

* dont block on head indexing for reading messages

* document why we're only checking for missing events for a single tipset

* document when we query for reverted events

* simplify event collection

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* fix test

* change event_id to id in the event table

* change head indexed timeout

* remove deprecated config options

* fail ETH RPC calls if ChainIndexer is disabled

* fix docs

* remove the tipset key cid func from lotus shed

* address review comments

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* chore(events): remove unnecessary DisableRealTimeFilterAPI (#12610)

* feat(cli): add --quiet to chainindex validate-backfill + cleanups (#12611)

* fix tests

* Apply suggestions from code review

Co-authored-by: Rod Vagg <[email protected]>

* error type for disabled chainindexer

* fix(chainindex): recompute tipset when we find no receipts

* fix(chainindexer): backfilling should halt when chain state data is missing and not backfill parents (#12619)

* fix backfilling UX

* Update chain/index/api.go

Co-authored-by: Rod Vagg <[email protected]>

* address review

---------

Co-authored-by: Rod Vagg <[email protected]>

* reduce log noise

* make jen

* make jen

* docs: finishing chain-indexer-overview-for-operators.md (#12600)

* Followup to PR #12450 for doc updates

This is being used to resolve the unresolved items in #12450 since that PR is unwieldly at this point.

* Incorporated some items and added TODOs based on unresolved items from #12450

* Incorporating more feedback

* Pointing to issue to learn about benefits

* Formatting fixes

* Apply most of the suggestions from @rvagg code review

Co-authored-by: Rod Vagg <[email protected]>

* Incorporating feedback from #12600 (comment)

* Addressing #12600 (comment) and more

* Moved chain-indexer docs to documentation
Renamed
Added ToC

We can move to lotus-docs later

* Update documentation/en/chain-indexer-overview-for-operators.md

Co-authored-by: Rod Vagg <[email protected]>

* Update documentation/en/chain-indexer-overview-for-operators.md

Co-authored-by: Rod Vagg <[email protected]>

* Added upgrade path when importing chain state from a snapshot.

* Typo fixes

* Update documentation/en/chain-indexer-overview-for-operators.md

Co-authored-by: Rod Vagg <[email protected]>

* chore(doc): "regular checks" section for chainindexer docs (#12612)

* Apply suggestions from @rvagg code review

Co-authored-by: Rod Vagg <[email protected]>

* Incorporating @aarshkshah1992 feedback

* Update documentation/en/chain-indexer-overview-for-operators.md

Co-authored-by: Rod Vagg <[email protected]>

---------

Co-authored-by: Rod Vagg <[email protected]>
Co-authored-by: Aarsh Shah <[email protected]>

* remove go mod replace

* remove unnecessary changes from CHANGELOG

* fix test

* compare events AMT root (#12632)

* fix(chainindex): retry transaction if database connection is lost (#12657)

* retry database lost connection

* log context cancellation

* address review

* fix gateway itest: no chainindexer for lite nodes

* fix changelog

---------

Co-authored-by: Rod Vagg <[email protected]>
Co-authored-by: Aryan Tikarya <[email protected]>
Co-authored-by: Steve Loeppky <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
skip/changelog This change does not require CHANGELOG.md update
Projects
Status: ☑️ Done (Archive)
Development

Successfully merging this pull request may close these issues.

2 participants