mirror of
https://github.com/penpot/penpot.git
synced 2026-10-30 14:30:37 +00:00
8b441be02f
`file_change` is the biggest relation of the database: 683 GB in production, of which almost all the TOAST is dead space (the legacy data is already migrated to `file_data`; what is left is the xlog of the revisions, 4.4 kB on average). It grows with every save of a file and every version, and its rows are spread over the whole table, so every query that filters by a file has to walk all of it. The table is now partitioned by `HASH (file_id)` with 16 partitions, the same bounds as `file_data`, so the queries that filter by a file can prune to one partition and the join between the two tables is resolved partition by partition. The primary key has to include the partition key, so it becomes `(file_id, id)`, and every access path of the table filters by `file_id` accordingly: the snapshot commands require it (2.19), the frontend sends it, and the garbage collector locks the pending rows by their `(file_id, id)` pair. The migration creates the partitioned table with the final schema and copies the rows, keeping the old one aside as `file_change_legacy` for the rollback. The copy happens inside the transaction of the migration, so it does not fit a big instance: there the migration is marked as applied and the procedure of `operations/` does it by hand, in phases, driven by the `updated_at` watermark. Along the way: - every mutation of the table moves `updated_at` to the moment of the mutation, so the phased copy does not lose the rows changed while it runs. It is never set to the `deleted_at` of the row, which for the system snapshots is a date in the future. - a new `kind` column tells the snapshots (1) and the revisions (2) of the table apart, and the two paths that insert in it fill it. It has no default on purpose, so an insert that forgets it fails instead of storing the wrong kind silently. It is a reserve: the queries still tell a snapshot from a revision by its label, and no constraint ties the two together. - the revision does not repeat the features of the file anymore: they were 348 bytes written on every save and nobody read them from there. - the legacy file data backend is gone, and with it the `data`, `data_ref_id` and `data_backend` columns, which nobody wrote anymore. - the migration copies the legacy blobs of the snapshots to `file_data` before the column goes away, and stops when a snapshot with legacy data is left without its blob: the backend is better down than losing content. - the foreign key of the legacy table on `file_id` goes away with the copy: it is `NO ACTION`, so while the legacy table has rows, the objects GC cannot delete the files they point to. - the quota of the snapshots of a team resolves the files of the team first and counts their versions with `= ANY`, instead of joining the team with every version of the table. - the pool sets `enable_partitionwise_join` on every connection, so the join between the two partitioned tables is resolved partition by partition. - the snapshot lookup reports an unknown snapshot with its own error instead of the generic "object not found" of the database helper, and the decode script fails instead of printing an empty JSON. - the procedure of `operations/` stops the backend and the workers during the cutover: the pool keeps server-side prepared statements, and a prepared statement holds the OID of the table it was prepared with, so a connection would keep writing to the legacy table after the rename. - the procedure creates the partitions with their final name, so the result is the schema a new installation gets, and its verification checks the names and not only how many there are. - the index of the system snapshots is gone: it filtered them by `deleted_at IS NULL`, and they always carry a deletion date in the future, so it never matched a row. - the configuration guide, the memories and the docstring of the decode script say that the `legacy-db` backend is gone, instead of offering a value that takes the service down. - a test fixes the schema the migration leaves: the 16 partitions with their name, the primary key, the foreign keys, the check of the kind, the six indexes, the absence of the legacy columns, and that an insert without a kind or with an unknown one fails. Closes #12416 AI-assisted-by: glm-5.3-flash AI-assisted-by: minimax-m3.1-flash
265 lines
12 KiB
Clojure
265 lines
12 KiB
Clojure
;; This Source Code Form is subject to the terms of the Mozilla Public
|
|
;; License, v. 2.0. If a copy of the MPL was not distributed with this
|
|
;; file, You can obtain one at http://mozilla.org/MPL/2.0/.
|
|
;;
|
|
;; Copyright (c) KALEIDOS INC Sucursal en España SL
|
|
|
|
(ns backend-tests.objects-gc-test
|
|
(:require
|
|
[app.common.logging :as l]
|
|
[app.common.time :as ct]
|
|
[app.config :as cf]
|
|
[app.db :as db]
|
|
[app.features.file-changes :as fchanges]
|
|
[app.rpc :as-alias rpc]
|
|
[app.storage :as sto]
|
|
[app.tasks.objects-gc :as ogc]
|
|
[backend-tests.helpers :as th]
|
|
[clojure.test :as t]
|
|
[mockery.core :refer [with-mocks]]))
|
|
|
|
(t/use-fixtures :once th/state-init)
|
|
(t/use-fixtures :each th/database-reset)
|
|
|
|
(defn- create-deleted-file!
|
|
"Creates a file and applies the deletion cascade on it (the same task the RPC
|
|
uses), so its children are marked as deleted for the given date. Returns the
|
|
file."
|
|
[profile i deleted-at]
|
|
(let [file (th/create-file* i {:profile-id (:id profile)
|
|
:project-id (:default-project-id profile)})]
|
|
(th/run-task! :delete-object {:object :file
|
|
:deleted-at deleted-at
|
|
:id (:id file)})
|
|
file))
|
|
|
|
(defn- block-file!
|
|
"Leaves an unmarked `file_media_object` pointing at the file, which is what
|
|
happens in production: a child that the deletion cascade did not mark. Its
|
|
foreign key towards `file` is `NO ACTION`, so it blocks the file deletion."
|
|
[file]
|
|
(let [media (sto/put-object! (::sto/storage th/*system*)
|
|
{::sto/content (sto/content "media")
|
|
:content-type "text/plain"
|
|
:bucket "file-media-object"})]
|
|
(th/create-file-media-object* {:file-id (:id file)
|
|
:media-id (:id media)})))
|
|
|
|
(defn- find-file
|
|
[id]
|
|
(th/db-get :file {:id id} {::db/remove-deleted false}))
|
|
|
|
(defn- chunk-error
|
|
"The first captured log record that reports a skipped chunk for `obj`,
|
|
waiting for it because logging is asynchronous."
|
|
[records obj]
|
|
(let [deadline (+ (System/currentTimeMillis) 5000)]
|
|
(loop []
|
|
(let [record (first (filter (fn [record]
|
|
(let [props (::l/props record)]
|
|
(and (= obj (get props :obj))
|
|
(= "error on chunk deletion, skipping it"
|
|
(get props :hint)))))
|
|
@records))]
|
|
(cond
|
|
record record
|
|
(> (System/currentTimeMillis) deadline) nil
|
|
:else (do (Thread/sleep 10) (recur)))))))
|
|
|
|
(t/deftest deletes-the-pending-files
|
|
(let [profile (th/create-profile* 1)
|
|
file (create-deleted-file! profile 1 (ct/in-past "10m"))
|
|
result (th/run-task! :objects-gc {})]
|
|
|
|
(t/is (pos? (:processed result)))
|
|
(t/is (nil? (find-file (:id file))))))
|
|
|
|
(t/deftest deletes-the-pending-file-changes
|
|
;; The pending file changes are locked and deleted by their (file-id, id)
|
|
;; pair, so the rows of a file that is pending deletion are gone after the
|
|
;; run.
|
|
(let [profile (th/create-profile* 1)
|
|
file (th/create-file* 1 {:profile-id (:id profile)
|
|
:project-id (:default-project-id profile)})
|
|
|
|
_ (th/db-insert! :file-change
|
|
{:file-id (:id file)
|
|
:kind fchanges/revision-kind
|
|
:deleted-at (ct/in-past "10m")})
|
|
|
|
result (th/run-task! :objects-gc {})]
|
|
|
|
(t/is (pos? (:processed result)))
|
|
(t/is (nil? (th/db-get :file-change {:file-id (:id file)}
|
|
{::db/remove-deleted false})))))
|
|
|
|
(t/deftest profile-cascade-keeps-profile-deletion-date
|
|
;; The cascade marks the owned rows with the deletion date of the profile,
|
|
;; like the RPC delete-profile flow does, not with the timestamp of the run.
|
|
;; The team survives because only the profiles chunk runs here, so its row
|
|
;; still shows that date.
|
|
(let [profile (th/create-profile* 1)
|
|
_ (th/db-update! :profile {:deleted-at (ct/in-past "30m")} {:id (:id profile)})
|
|
|
|
deleted-at (:deleted-at (th/db-get :profile {:id (:id profile)}
|
|
{::db/remove-deleted false}))
|
|
|
|
cfg {::db/pool th/*pool*
|
|
::sto/storage (::sto/storage th/*system*)
|
|
::ogc/timestamp deleted-at}
|
|
total (#'ogc/delete-profiles-chunk cfg [(:id profile)])]
|
|
|
|
(t/testing "the profile is gone and the chunk reports it"
|
|
(t/is (= 1 total))
|
|
(t/is (nil? (th/db-get :profile {:id (:id profile)} {::db/remove-deleted false}))))
|
|
|
|
(t/testing "the owned team keeps the deletion date of the profile"
|
|
(t/is (= deleted-at (:deleted-at (th/db-get :team {:id (:default-team-id profile)}
|
|
{::db/remove-deleted false})))))))
|
|
|
|
(t/deftest blocked-chunk-does-not-stop-the-run
|
|
;; Chunks of one file: the blocked file takes down its own chunk and nothing
|
|
;; else, so the clean file of the next chunk is still deleted and the run
|
|
;; finishes. The next run retries the blocked file.
|
|
(let [profile (th/create-profile* 1)
|
|
blocked (create-deleted-file! profile 1 (ct/in-past "30m"))
|
|
clean (create-deleted-file! profile 2 (ct/in-past "20m"))
|
|
|
|
_ (block-file! blocked)
|
|
|
|
records (atom [])
|
|
_ (add-watch l/log-record ::capture
|
|
(fn [_ _ _ record] (swap! records conj record)))
|
|
result (try
|
|
(th/run-task! :objects-gc {:chunk-size 1})
|
|
(finally
|
|
(remove-watch l/log-record ::capture)))]
|
|
|
|
(t/testing "the clean file of the next chunk is deleted"
|
|
(t/is (nil? (find-file (:id clean)))))
|
|
|
|
(t/testing "the blocked file is still pending"
|
|
(t/is (some? (find-file (:id blocked)))))
|
|
|
|
(t/testing "the failed chunk is reported"
|
|
(t/is (some? (chunk-error records "file"))))
|
|
|
|
(t/testing "the run finishes"
|
|
(t/is (pos? (:processed result))))))
|
|
|
|
(defn- submitted-events
|
|
[audit-mock]
|
|
(->> (:call-args-list @audit-mock)
|
|
(map second)))
|
|
|
|
(defn- events-named
|
|
[audit-mock event-name]
|
|
(filter #(= event-name (:name %))
|
|
(submitted-events audit-mock)))
|
|
|
|
(t/deftest gc-emits-hard-delete-objects-for-teams
|
|
;; When the GC physically deletes a team, the projection must drop its row,
|
|
;; so the event carries the purged ids.
|
|
(with-mocks [audit-mock {:target 'app.loggers.audit/submit :return nil}]
|
|
(with-redefs [cf/flags (conj cf/flags :audit-log)]
|
|
(let [profile (th/create-profile* 1 {:is-active true})
|
|
team (th/create-team* 71 {:profile-id (:id profile)})]
|
|
(th/command! {::th/type :delete-team
|
|
::rpc/profile-id (:id profile)
|
|
:id (:id team)})
|
|
;; Run the queued cascade so the default project is marked deleted too,
|
|
;; then purge everything the way production does.
|
|
(th/run-pending-tasks!)
|
|
(th/reset-mock! audit-mock)
|
|
(th/run-task! :objects-gc {:skip-delay true})
|
|
(t/is (nil? (th/db-get :team {:id (:id team)} {::db/remove-deleted false})))
|
|
(let [events (filter #(contains? (:props %) :team-ids)
|
|
(events-named audit-mock "hard-delete-objects"))]
|
|
(t/is (= 1 (count events)))
|
|
(let [event (first events)]
|
|
(t/is (= "command" (:type event)))
|
|
(t/is (= [(:id team)] (get-in event [:props :team-ids])))
|
|
(t/is (= "objects-gc" (get-in event [:context :triggered-by])))))))))
|
|
|
|
(t/deftest gc-emits-hard-delete-objects-for-projects
|
|
(with-mocks [audit-mock {:target 'app.loggers.audit/submit :return nil}]
|
|
(with-redefs [cf/flags (conj cf/flags :audit-log)]
|
|
(let [profile (th/create-profile* 1 {:is-active true})
|
|
team (th/create-team* 72 {:profile-id (:id profile)})
|
|
project (th/create-project* 72 {:profile-id (:id profile)
|
|
:team-id (:id team)})]
|
|
(th/command! {::th/type :delete-project
|
|
::rpc/profile-id (:id profile)
|
|
:id (:id project)})
|
|
(th/run-pending-tasks!)
|
|
(th/reset-mock! audit-mock)
|
|
(th/run-task! :objects-gc {:skip-delay true})
|
|
(t/is (nil? (th/db-get :project {:id (:id project)} {::db/remove-deleted false})))
|
|
(let [events (filter #(contains? (:props %) :project-ids)
|
|
(events-named audit-mock "hard-delete-objects"))]
|
|
(t/is (= 1 (count events)))
|
|
(let [event (first events)]
|
|
(t/is (= "command" (:type event)))
|
|
(t/is (= [(:id project)] (get-in event [:props :project-ids])))
|
|
(t/is (= "objects-gc" (get-in event [:context :triggered-by])))))))))
|
|
|
|
(t/deftest gc-emits-hard-delete-objects-for-files
|
|
(with-mocks [audit-mock {:target 'app.loggers.audit/submit :return nil}]
|
|
(with-redefs [cf/flags (conj cf/flags :audit-log)]
|
|
(let [profile (th/create-profile* 1 {:is-active true})
|
|
team (th/create-team* 73 {:profile-id (:id profile)})
|
|
project (th/create-project* 73 {:profile-id (:id profile)
|
|
:team-id (:id team)})
|
|
file (th/create-file* 73 {:profile-id (:id profile)
|
|
:project-id (:id project)})]
|
|
(th/command! {::th/type :delete-file
|
|
::rpc/profile-id (:id profile)
|
|
:id (:id file)})
|
|
(th/run-pending-tasks!)
|
|
(th/reset-mock! audit-mock)
|
|
(th/run-task! :objects-gc {:skip-delay true})
|
|
(t/is (nil? (th/db-get :file {:id (:id file)} {::db/remove-deleted false})))
|
|
(let [events (filter #(contains? (:props %) :file-ids)
|
|
(events-named audit-mock "hard-delete-objects"))]
|
|
(t/is (= 1 (count events)))
|
|
(let [event (first events)]
|
|
(t/is (= "command" (:type event)))
|
|
(t/is (= [(:id file)] (get-in event [:props :file-ids])))
|
|
(t/is (= "objects-gc" (get-in event [:context :triggered-by])))))))))
|
|
|
|
(t/deftest gc-emits-no-event-without-audit-log-flag
|
|
;; The GC must stay silent when the audit log is off, even though the row is
|
|
;; still purged.
|
|
(with-mocks [audit-mock {:target 'app.loggers.audit/submit :return nil}]
|
|
(let [profile (th/create-profile* 1 {:is-active true})
|
|
team (th/create-team* 74 {:profile-id (:id profile)})]
|
|
(th/command! {::th/type :delete-team
|
|
::rpc/profile-id (:id profile)
|
|
:id (:id team)})
|
|
(th/run-pending-tasks!)
|
|
(th/reset-mock! audit-mock)
|
|
(th/run-task! :objects-gc {:skip-delay true})
|
|
(t/is (nil? (th/db-get :team {:id (:id team)} {::db/remove-deleted false})))
|
|
(t/is (empty? (:call-args-list @audit-mock))))))
|
|
|
|
(t/deftest gc-emits-events-only-for-committed-chunks
|
|
(with-mocks [audit-mock {:target 'app.loggers.audit/submit :return nil}]
|
|
(with-redefs [cf/flags (conj cf/flags :audit-log)]
|
|
(let [profile (th/create-profile* 1)
|
|
rolled-back (create-deleted-file! profile 1 (ct/in-past "30m"))
|
|
blocked (create-deleted-file! profile 2 (ct/in-past "20m"))
|
|
clean (create-deleted-file! profile 3 (ct/in-past "10m"))]
|
|
(block-file! blocked)
|
|
(th/reset-mock! audit-mock)
|
|
(let [result (th/run-task! :objects-gc {:chunk-size 2})
|
|
events (filter #(contains? (:props %) :file-ids)
|
|
(events-named audit-mock "hard-delete-objects"))]
|
|
(t/testing "the failed chunk keeps both files"
|
|
(t/is (some? (find-file (:id rolled-back))))
|
|
(t/is (some? (find-file (:id blocked)))))
|
|
(t/testing "the next chunk commits and reports only its deleted file"
|
|
(t/is (nil? (find-file (:id clean))))
|
|
(t/is (pos? (:processed result)))
|
|
(t/is (= 1 (count events)))
|
|
(t/is (= [(:id clean)] (get-in (first events) [:props :file-ids])))))))))
|