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

feat: add v1 schema #528

Merged
merged 1 commit into from
Jun 10, 2021
Merged

feat: add v1 schema #528

merged 1 commit into from
Jun 10, 2021

Conversation

iand
Copy link
Contributor

@iand iand commented May 26, 2021

Consolidate all v0 migrations into a new v1 schema with some improvements. Visor supports both v0 and v1 schemas when writing directly to database so this won't block feature updates while we migrate to the newer schema. CSV output uses the latest schema.

Incorporates the following PRs:

Also:

NOTE: I did look at changing the schema name to visor (for #218) but it turned out to be a much larger and more invasive change than I wanted, especially since I am trying to give us a period of compatibility to allow us to migrate between v0 and v1 of the schema. Go-pg has no explicit support for different schemas (workaround here) and there are some rough edges when using some of the timescale functions such as set_integer_now_func (which needs the schema embedded in the name of the table passed to it)

Summary of Schema Changes

Added a new function current_height which returns the expected current epoch. All hypertables have been set to use current_height as their "now" function.

The following tables are no longer hypertables due to low rates of growth:

  • chain_powers
  • chain_rewards
  • miner_infos
  • miner_sector_deals
  • multisig_approvals

The following tables have been converted to hypertables:

  • derived_gas_outputs: weekly chunks

The following tables have had their hypertable chunking adjusted:

  • messages: weekly, was daily
  • receipts: weekly, was daily
  • miner_sector_posts: daily, was weekly

The following tables have additional columns:

  • chain_economics: height, fil_reserve_disbursed
  • id_addresses: height
  • derived_gas_outputs: actor_family

The following tables have had columns changed to use numeric data type:

  • chain_economics: circulating_fil, vested_fil, mined_fil, burnt_fil, locked_fil
  • chain_powers: total_raw_bytes_power, total_raw_bytes_committed, total_qa_bytes_power, total_qa_bytes_committed, total_pledge_collateral, qa_smoothed_position_estimate, qa_smoothed_velocity_estimate
  • chain_rewards: cum_sum_baseline, cum_sum_realized, effective_baseline_power, new_baseline_power, new_reward_smoothed_position_estimate, new_reward_smoothed_velocity_estimate, total_mined_reward, new_reward numeric
  • derived_gas_outputs: value, gas_fee_cap, gas_premium, parent_base_fee, base_fee_burn, over_estimation_burn, miner_penalty, miner_tip numeric, refund
  • message_gas_economy: gas_limit_total, gas_limit_unique_total, base_fee, value, gas_fee_cap, gas_premium, fee_debt
  • miner_locked_funds: locked_funds, initial_pledge, pre_commit_deposits
  • miner_pre_commit_infos: pre_commit_deposit, deal_weight, verified_deal_weight
  • miner_sector_infos: deal_weight, verified_deal_weight, initial_pledge, expected_day_reward, expected_storage_pledge
  • parsed_messages: value
  • power_actor_claims: raw_byte_power, quality_adj_power

New tables:

  • internal_messages
  • internal_parsed_messages

@iand iand force-pushed the iand/schema-v1-2 branch from ccb9363 to 0111eba Compare May 26, 2021 10:25
@codecov-commenter
Copy link

codecov-commenter commented May 26, 2021

Codecov Report

Merging #528 (254274c) into master (8b0faf2) will decrease coverage by 1.5%.
The diff coverage is 14.6%.

@@           Coverage Diff            @@
##           master    #528     +/-   ##
========================================
- Coverage    42.1%   40.6%   -1.6%     
========================================
  Files          40      40             
  Lines        3100    3237    +137     
========================================
+ Hits         1308    1315      +7     
- Misses       1583    1713    +130     
  Partials      209     209             

@iand
Copy link
Contributor Author

iand commented May 28, 2021

Since this is a wide reaching PR I requested review from all 3 of you (@hsanjuan, @frrist, @placer14 ) to bring it to your attention, not necessarily to review it in its entirety.

I expect we will tag visor as v0.7.0 after this is merged.

Copy link
Member

@frrist frrist left a comment

Choose a reason for hiding this comment

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

still reviewing, adding a comment while I remember.

NewRewardSmoothedPositionEstimate string `pg:"type:numeric,notnull"`
NewRewardSmoothedVelocityEstimate string `pg:"type:numeric,notnull"`
TotalMinedReward string `pg:"type:numeric,notnull"`
NewReward string `pg:"type:numeric,use_zero"`
Copy link
Member

@frrist frrist May 27, 2021

Choose a reason for hiding this comment

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

This should be notnull instead of use_zero, since its a string

Copy link
Contributor Author

Choose a reason for hiding this comment

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

fixed

@hsanjuan
Copy link
Contributor

I did look at changing the schema name to visor (for #218) but it turned out to be a much larger and more invasive change than I wanted

Can we allow a configurable single schema name to use? When we load the backfill it should preferably go into the visor schema on a pristine DB, so compatibility should not matter. We can keep backfilling on that DB until Lily can take over.

@iand
Copy link
Contributor Author

iand commented Jun 1, 2021

I did look at changing the schema name to visor (for #218) but it turned out to be a much larger and more invasive change than I wanted

Can we allow a configurable single schema name to use? When we load the backfill it should preferably go into the visor schema on a pristine DB, so compatibility should not matter. We can keep backfilling on that DB until Lily can take over.

See this commit d6273b1

@@ -42,5 +77,16 @@ func (ml MinerFeeDebtList) Persist(ctx context.Context, s model.StorageBatch, ve
if len(ml) == 0 {
return nil
}

if version.Major != 1 {
// Support older versions, but in a non-optimal way
Copy link
Member

Choose a reason for hiding this comment

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

Is something like this worth logging? (Here and elsewhere in the persist methods)

Copy link
Contributor Author

Choose a reason for hiding this comment

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

It could be a really noisy log messages since it would fire on every write. I'll look at issuing a warning when visor starts

if target.Major != dbVersion.Major {
// Version 0.0 indicates that the database has not been initialised with a schema yet
if target.Major != dbVersion.Major &&
!(dbVersion.Major == 0 && dbVersion.Patch == 0) {
Copy link
Member

Choose a reason for hiding this comment

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

nit: if this check is performed elsewhere dbVersion.IsInitalized() would be a bit cleaner.

Copy link
Contributor Author

Choose a reason for hiding this comment

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

I rewrote this area to check explicitly whether the version tables are initialized or not

@iand
Copy link
Contributor Author

iand commented Jun 3, 2021

Added actor_family column to derived_gas_outputs which is just the last segment of the actor name after the version but having it in the table allows us to index it and remove various wildcard hacks that are used in queries and monitoring.

@iand
Copy link
Contributor Author

iand commented Jun 3, 2021

Added internal message tables as per #482 (comment)

From string `pg:",notnull"`
To string `pg:",notnull"`
Value string `pg:"type:numeric,notnull"`
Method uint64 `pg:",use_zero"`
Copy link
Member

Choose a reason for hiding this comment

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

should be of type string based on comment here: #482 (comment)

Copy link
Contributor Author

Choose a reason for hiding this comment

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

good catch, will fix

Copy link
Member

Choose a reason for hiding this comment

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

I've copy-pasted these structs (I fixed this bit) into my internal_message branch: #533. Are you alright with #533 landing as a follow-on to this?

Copy link
Contributor Author

Choose a reason for hiding this comment

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

Yep, let's do it as a follow-up

@iand iand force-pushed the iand/schema-v1-2 branch 3 times, most recently from 6452539 to 32306a9 Compare June 10, 2021 13:12
* feat: enable models to be version aware

* Use a structured version

* Move schema migrations to new versioned package

* Add base SQL for schemas and update README

* Fix some spelling errors caught by lint

* Appease the lint god

* Incorporate pr review feedback

Create v1 schema and adjust hypertable parameters

Use numeric type to represent numbers

add integer now function to each hypertable

multisig tx param field is nullable

Add Height to IDAddress and ChainEconomics

Fix ups to ensure schema v1 matches models

Some lint fixes

Ensure older schema version can be installed in new database

Add add FilReserveDisbursed for chain economics

Support backwards compatibility with v0 models

Fix genesis test

Support custom postgresql schema names

Explicitly check whether database has initialized version tables

Fix ChainReward.NewReward to be notnull

Convert value column in parsed_messages to numeric

Add actor_family colum to derived_gas_outputs

Add internal_messages and internal_parsed_messages models

Make InternalParsedMessage.Method a string to match schema

Remove extension comment

Only set base schema search path if non-default

Fix typo

Add version number to initial version tables

Fix template typo

Revert adding version number to initial version tables

Remove extension creation

Remove owner statements from v1 schema

Use fixed statediff

Use merged version in statediff repo

tweak logger name
@iand iand force-pushed the iand/schema-v1-2 branch from 32306a9 to 28961a0 Compare June 10, 2021 13:12
@iand iand self-assigned this Jun 10, 2021
@iand iand merged commit ddacfc7 into master Jun 10, 2021
@iand iand deleted the iand/schema-v1-2 branch June 10, 2021 17:44
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
4 participants