Skip to content
This repository has been archived by the owner on Jul 24, 2021. It is now read-only.

Releases: joyent/conch-api

v2.33.4

12 Aug 18:34
9b5571c
Compare
Choose a tag to compare

Release v2.33.4

PR #863 - v2: log latency

  • Branch: ether/log-latency
  • Revision: 9b5571c
  • Labels: enhancement, v2

also log the request duration, in milliseconds

as per guidelines in joyent/eng:docs/index.md

(cherry-picked from v3)

People

Commits

  • 8b7e05 - do not set response code on intentional rollback
  • 5f55df - also log the request duration, in milliseconds

v2.33.3

09 Aug 16:48
Compare
Choose a tag to compare
v2.33.3 Pre-release
Pre-release

Release v2.33.3

PR #858 - well, that was a bust...

  • Branch: ether/v2-well-that-was-a-bust
  • Revision: 9b44fcf
  • Labels: database, v2

Back out of most of the work done in 2.33.2, as restructuring validation_result and validation_state_member took too long and resulted in circular deadlocks if a report was processed at the same time.

Additionally, the uuid-ossp extension could not be dropped at the user level run in migrations.

The only thing we can keep is removing the use of the uuid-ossp extension in the tables that were still doing so.

People

Commits

  • 66e0cc - Revert "move result_order to validation_state_result"
  • 29f158 - back out of the dropping of the postgres extension

v2.33.2

08 Aug 20:17
3d7b91e
Compare
Choose a tag to compare
v2.33.2 Pre-release
Pre-release

Release v2.33.2

Achtung!! Do not deploy this version. The majority of its changes were reverted in v2.33.3.

PR #855 - v2: collapse validation result more

  • Branch: ether/v2-collapse-validation_result-more
  • Revision: 6e641e6
  • Labels: database, device reports, v2

move result_order to validation_state_result

This makes validation_result rows more reusable across multiple validation_states.
Side effect:

  • result_order is now incremented across all validation_results for a state,
    not just across results for a single validation

This will allow conch merge_validation_results to collapse even more rows
together than it could before.

People

Commits

  • 268ff0 - add many-to-many bridge accessors for join tables with extra columns
  • 4e987e - fix CODEOWNERS, as done in master for v3
  • 544ee6 - move result_order to validation_state_result

PR #856 - drop all uses of the unnecessary uuid-ossp extension

  • Branch: ether/v2-drop-uuid-ossp
  • Revision: 3d7b91e
  • Labels: database, device reports, v2

See the last section of https://www.postgresql.org/docs/9.6/uuid-ossp.html

(cherry picked from commit 6272a7a, from v3
and migration 102.)

People

Commits

  • 5c9ebc - drop all uses of the unnecessary uuid-ossp extension

v2.33.1

08 Aug 16:37
85b727c
Compare
Choose a tag to compare
v2.33.1 Pre-release
Pre-release

Release v2.33.1

PR #854 - go through pages in reverse

  • Branch: ether/v2-collapse-validation_result-fix
  • Revision: 85b727c
  • Labels: v2, validation

..so we don't have to run the script multiple times to find all the groups

People

Commits

  • 8b40a8 - go through pages in reverse

v2.33.0

07 Aug 21:51
8521644
Compare
Choose a tag to compare
v2.33.0 Pre-release
Pre-release

Release v2.33.0

PR #853 - v2 script to collapse validation_result rows together with "identical" data

  • Branch: ether/v2-collapse-validation_result
  • Revision: 8521644
  • Labels: device reports, high priority, v2, validation

This is projected to result in a 96% reduction in validation_result rows on production.

People

Commits

  • 7f383b - v2 script to collapse validation_result rows together with "identical" data

v2.32.5

02 Aug 20:33
bc562fe
Compare
Choose a tag to compare

Release v2.32.5

PR #851 - v2: fix slow validation result inserts

  • Branch: ether/fix-slow-validation_result-inserts
  • Revision: bc562fe
  • Labels: breaking, database, v2, validation

People

Commits

  • a4e3ca - fix check for log line
  • 40852c - speed up validation plan runs by adding an appropriate index

v2.32.4

02 Aug 18:10
ab7287c
Compare
Choose a tag to compare
v2.32.4 Pre-release
Pre-release

Release v2.32.4

PR #848 - v2: undef message 3

  • Branch: ether/v2-undef-message-3
  • Revision: 83a9aaa
  • Labels: bug, v2, validation, wat

For errors with \n inside them, the filename and line number do not appear on the first line

People

Commits

  • 6a1968 - keep $message defined by parsing it properly

PR #850 - v2: more report logging

  • Branch: ether/more-report-logging
  • Revision: ab7287c
  • Labels:

..for debugging an outstanding prod issue

People

Commits

  • 7db931 - add a bit more debug logging for report process

v2.32.3

30 Jul 19:33
f19f0f6
Compare
Choose a tag to compare

Release v2.32.3

PR #847 - $message is undef here too

  • Branch: ether/v2-undef-message-2
  • Revision: f19f0f6
  • Labels: bug, v2, validation

People

Commits

  • 5b25e3 - $message is undef here too

v2.32.2

30 Jul 19:18
3bd5087
Compare
Choose a tag to compare
v2.32.2 Pre-release
Pre-release

Release v2.32.2

PR #844 - v2: log unmatched routes

  • Branch: ether/log-unmatched-routes
  • Revision: 444d57e
  • Labels: api, v2

This will make it easier to detect outdated clients that are using invalid URLs

(cherry picked from commit 484967a)

People

Commits

  • 23d9b0 - log an error when no matching route is found for a URL

PR #845 - fix: hdd_size validation when specification is unset

  • Branch: ether/v2-fix-hardware_product-specification
  • Revision: 5e6389c
  • Labels: bug, v2, validation

This fixes: "malformed JSON string..." in null hardware_product specification entries

(cherry picked from commit ed9282e)

People

Commits

  • 1af2dc - fix: hdd_size validation when specification is unset

PR #846 - $message should never be undef here

  • Branch: ether/v2-undef-message
  • Revision: 3bd5087
  • Labels: bug, v2, validation, wat

People

Commits

  • c66ffc - $message should never be undef here

v3.0.0-a3

29 Jul 21:13
bcfcb10
Compare
Choose a tag to compare
v3.0.0-a3 Pre-release
Pre-release

Release v3.0.0-a3

Warning -- one of the migrations in this release is broken. See commit ff16ef6 for the fix.

PR #838 - In Docker, upgrade Postgres to v10

  • Branch: sungo/docker-pg-10
  • Revision: 9932b8d
  • Labels: database, v3

Both dockerfiles now use postgres 10 which ships with ubuntu:bionic.

Part of https://github.com/joyent/buildops-infra/issues/87

People

Commits

  • 2699ff - the docker-compose file drops a "postgres" directory in the root. We dont want docker to try and pull that into images
  • b34899 - upgrade the dockerfiles to use postgres 10 which ships with ubuntu:bionic
  • 9faa90 - increase the database version warning to 10.9, the version the ubuntu 18.04 LTS currently ships

PR #836 - v3: more small fixes

  • Branch: ether/v3-more-small-fixes
  • Revision: b4fcf46
  • Labels: api, breaking, bug, database, tests

some GHIs are closed:

  • fix: allow setting asset_tag and serial_number in POST /rack (closes #835)
  • remove GET /workspace/:workspace_id_or_name/rack/:rack_id (closes #796)
  • fix: allow setting of device asset_tag to null (closes again #668)

..and a few small changes to some json schemas for consistency and correctness

People

Commits

  • 6c1120 - fix: allow setting of device asset_tag to null
  • 52fac1 - json schemas: tighten up string types so empty string is disallowed in most inputs
  • d8e488 - remove GET /workspace/:workspace_id_or_name/rack/:rack_id
  • 27c41c - update all dependencies to latest versions
  • 23233c - when resetting a password, ensure it will fail
  • e5ac25 - fix bad variable name in error message
  • f2943c - show the log entries when log tests fail
  • 701009 - fix: allow setting asset_tag and serial_number in POST /rack
  • 14b172 - add user_account.last_seen, and update it every time the user authenticates
  • e2fa15 - fix precedence issue in codeowners
  • 76bd01 - efficiency: extra try/catch features are not used here
  • 4f1036 - NYTProf is not normally needed, and does not install in perl 5.31.1
  • 3ff6c7 - tighter whitespace config, and fixes
  • cf4cc3 - json schemas, breaking: normalize "parent_id" -> "parent_workspace_id"
  • d295cc - clean up .gitignore
  • a0afe4 - fix: use secrets consistently

PR #839 - v3: more small fixes, part 2

  • Branch: ether/v3-more-small-fixes-part2
  • Revision: 7f006bf
  • Labels: database, enhancement, refactor, tests
  • some db queries have been rewritten for speed
  • atomic swapping of device serial numbers and locations is now supported, through the use of deferred constraints
  • multiple datacenters with the same fields are no longer permitted

People

Commits

  • 61e5f1 - rewrite this query to remove "..where device.id not in (HUGE LIST)"
  • e5736b - eliminate needless subselects in delete queries
  • 4e63b9 - simplify this query
  • 9f8a83 - accomodate atomically swapping device serials
  • 4be12b - also provide for atomic swapping of device locations
  • 1518f6 - convert null_user fixture to a generated definition
  • 8fd18a - WorkspaceUser: speed up this query by not inflating to objects
  • b290ab - use already-stashed workspace name, when available
  • 04b2d6 - prevent multiple datacenters from being created with the same not-nullable fields
  • 14bfac - rename this resultset method to be more clear
  • a725bf - drop support for legacy database configs
  • 64eb97 - also share fixtures between Test::Conch instances
  • ec4f3a - also log the name and email of the authenticating user
  • 3b4f4a - dump unexpected mail(s) sent
  • 99a377 - use proper response code for auth failures
  • b9a9f4 - clean up a bit more test cruft
  • 55c2a0 - fix bad documentation
  • 7954e0 - sort results in response to GET /user
  • e0ebfa - better docs re db credentials
  • 917f1e - test fetching a deactivated user
  • af6f43 - add link from main docs page to module index
  • f32b70 - find_user no longer needs to be in its own helper
  • 5eb3a9 - log who just logged in
  • 19b85f - fix: allow browsers that don't pass X-Conch-UI

PR #841 - do not dump the entire device report into the log on error

  • Branch: ether/v3-no-log-report-data
  • Revision: bcfcb10
  • Labels: device reports, v2, validation

we already logged the device_report id, and we include the device id here, so
that is sufficient to find the bad data

(same as PR#840, but for v3)

People

Commits

  • 36fd37 - do not dump the entire device report into the log on error