ui: reroute new db page to legacy page if version error occurs #133843

xinhaoz · 2024-10-30T14:42:08Z

A version incompatibility error is returned when the migration to create the system.table_metadata table and update job aren't yet completed. On the UI we should reroute to the legacy page when this occurs.

Epic: CRDB-37558

Jira issue: CRDB-43791

The migrations that support the v2 db page runs on finalization. Rather than show a non-functional page in an unfinalized cluster state let's reroute to the legacy page. Epic: CRDB-37558 Fixes: cockroachdb#133843 Release note (ui change): The v2 db pages will only be available post ugprade finalization to 24.3. Prior to that we'll continue to show the legacy page when the cluster is in its unfinalized state.

133895: ui: reroute v2 db page to legacy page when cluster is unfinalized r=xinhaoz a=xinhaoz The migrations that support the v2 db page runs on finalization. Rather than show a non-functional page in an unfinalized cluster state let's reroute to the legacy page. Epic: CRDB-37558 Fixes: #133843 Release note (ui change): The v2 db pages will only be available post ugprade finalization to 24.3. Prior to that we'll continue to show the legacy page when the cluster is in its unfinalized state. Co-authored-by: Xin Hao Zhang <[email protected]>

132877: kv: deflake TestLeaseQueueLeasePreferencePurgatoryError r=nvanbenschoten a=nvanbenschoten Fixes #132652. This commit fixes a flake in TestLeaseQueueLeasePreferencePurgatoryError where a delayed span config split could trip up the test and cause it to flake. We now wait for the span config split to be processed before waiting on the lease queue's purgatory, which should eliminate the flake. Release note: None 133568: changefeedccl: add PTS to system.users r=andyyang890 a=asg0451 Add `system.users` to the list of system tables that changefeeds protect with PTS. This table is required for CDC Queries. Part of: #128806 Release note (enterprise change): Add `system.users` to the list of system tables that changefeeds protect with PTS. This table is required for CDC Queries. 133873: raft: ensure leaderMaxSupported does not regress when bumping terms r=nvanbenschoten a=arulajmani Previously, it was possible for a leader to regress leaderMaxSupported by calling an election at a higher term. This was because we weren't careful to recognize this case on the leader, and followers had special case handling that allowed a leader to override `inFortifyLease`. Together, this could cause lease regressions for LeaderLeases in some rare cases. This patch fixes this issue by removing special case handling in `inFortifyLease`. We also remove a special case in the handling of `MsgHup` which allowed a leader to step down and campaign without de-fortifying itself. Fixes #133764 Release note: None 133895: ui: reroute v2 db page to legacy page when cluster is unfinalized r=xinhaoz a=xinhaoz The migrations that support the v2 db page runs on finalization. Rather than show a non-functional page in an unfinalized cluster state let's reroute to the legacy page. Epic: CRDB-37558 Fixes: #133843 Release note (ui change): The v2 db pages will only be available post ugprade finalization to 24.3. Prior to that we'll continue to show the legacy page when the cluster is in its unfinalized state. 134082: cli: improve tsdump upload concurrency behavior r=arjunmahishi a=dhartunian - wrap concurrent error collection and status printing in necessary mutexes. - reduce goroutine spawning to 10 for primary loop which avoids rate limit errors from datadog. Release note: None 134216: sql/catalog: avoid multiple interface boxing allocs from `descpb.NameInfo` r=nvanbenschoten a=nvanbenschoten When we cast a `descpb.NameInfo` to a `catalog.NameKey`, we incur at least one heap allocation to box the struct in an interface. This commit ensures that it's only one, and not multiple. The change was motivated by observing two heap allocations in `lookupStoreCacheID`. Ideally, we would have no heap allocations here, but this is a step in the right direction. While here, we also clean up a few similar cases. ``` name old time/op new time/op delta Sysbench/SQL/oltp_read_only 36.3ms ± 4% 36.0ms ± 3% ~ (p=0.730 n=9+9) Sysbench/SQL/oltp_write_only 19.3ms ± 6% 18.8ms ± 4% -2.36% (p=0.043 n=9+10) Sysbench/SQL/oltp_read_write 58.7ms ± 3% 57.2ms ± 2% -2.49% (p=0.000 n=10+10) Sysbench/SQL/oltp_point_select 1.61ms ± 1% 1.56ms ± 2% -2.59% (p=0.000 n=9+10) Sysbench/SQL/oltp_begin_commit 594µs ± 4% 608µs ± 7% ~ (p=0.089 n=10+10) name old alloc/op new alloc/op delta Sysbench/SQL/oltp_read_only 1.05MB ± 1% 1.04MB ± 0% -0.64% (p=0.004 n=9+10) Sysbench/SQL/oltp_write_only 405kB ± 1% 404kB ± 1% ~ (p=0.720 n=9+10) Sysbench/SQL/oltp_read_write 1.50MB ± 1% 1.51MB ± 1% ~ (p=0.447 n=9+10) Sysbench/SQL/oltp_point_select 35.3kB ± 2% 35.1kB ± 2% ~ (p=0.278 n=9+10) Sysbench/SQL/oltp_begin_commit 18.9kB ± 0% 18.9kB ± 1% ~ (p=0.055 n=10+10) name old allocs/op new allocs/op delta Sysbench/SQL/oltp_read_only 6.96k ± 1% 6.86k ± 1% -1.37% (p=0.000 n=9+10) Sysbench/SQL/oltp_write_only 3.62k ± 1% 3.58k ± 0% -0.96% (p=0.000 n=9+10) Sysbench/SQL/oltp_read_write 11.0k ± 1% 10.9k ± 1% -0.56% (p=0.030 n=9+10) Sysbench/SQL/oltp_point_select 317 ± 1% 312 ± 0% -1.74% (p=0.000 n=9+10) Sysbench/SQL/oltp_begin_commit 140 ± 0% 140 ± 0% ~ (p=1.000 n=10+10) ``` Epic: None Release note: None 134217: sql: fix recently introduced data race r=yuzefovich a=yuzefovich Recently, in a5b8d06 we introduced a possible data race. Namely, that change modifies a proto inside of the physical plan to cap the slice of types during the vectorized operator planning in order to prevent a type schema corruption due to slice aliasing. However, it's been assumed in a couple of places that the physical plan is immutable. This commit clarifies that the physical plan is immutable after its finalization and moves the capping into the finalization. Fixes: #133934. Release note: None Co-authored-by: Nathan VanBenschoten <[email protected]> Co-authored-by: Miles Frankel <[email protected]> Co-authored-by: Arul Ajmani <[email protected]> Co-authored-by: Xin Hao Zhang <[email protected]> Co-authored-by: David Hartunian <[email protected]> Co-authored-by: Yahor Yuzefovich <[email protected]>

133895: ui: reroute v2 db page to legacy page when cluster is unfinalized r=xinhaoz a=xinhaoz The migrations that support the v2 db page runs on finalization. Rather than show a non-functional page in an unfinalized cluster state let's reroute to the legacy page. Epic: CRDB-37558 Fixes: #133843 Release note (ui change): The v2 db pages will only be available post ugprade finalization to 24.3. Prior to that we'll continue to show the legacy page when the cluster is in its unfinalized state. 134082: cli: improve tsdump upload concurrency behavior r=arjunmahishi a=dhartunian - wrap concurrent error collection and status printing in necessary mutexes. - reduce goroutine spawning to 10 for primary loop which avoids rate limit errors from datadog. Release note: None 134154: raft: enable store liveness in raft unit tests phase 3 r=iskettaneh a=iskettaneh The first 2 commits are based on #133971 This commit modifies the following tests to allow them to run in both store liveness enabled/disabled configurations: 1) TestDisruptiveFollower 2) TestDisruptiveFollowerPreVote 3) TestDuelingCandidates 4) TestDuelingPreCandidates 5) TestFreeStuckCandidateWithCheckQuorum 6) TestLeaderElectionWithCheckQuorum References: #132241 Release note: None 134216: sql/catalog: avoid multiple interface boxing allocs from `descpb.NameInfo` r=nvanbenschoten a=nvanbenschoten When we cast a `descpb.NameInfo` to a `catalog.NameKey`, we incur at least one heap allocation to box the struct in an interface. This commit ensures that it's only one, and not multiple. The change was motivated by observing two heap allocations in `lookupStoreCacheID`. Ideally, we would have no heap allocations here, but this is a step in the right direction. While here, we also clean up a few similar cases. ``` name old time/op new time/op delta Sysbench/SQL/oltp_read_only 36.3ms ± 4% 36.0ms ± 3% ~ (p=0.730 n=9+9) Sysbench/SQL/oltp_write_only 19.3ms ± 6% 18.8ms ± 4% -2.36% (p=0.043 n=9+10) Sysbench/SQL/oltp_read_write 58.7ms ± 3% 57.2ms ± 2% -2.49% (p=0.000 n=10+10) Sysbench/SQL/oltp_point_select 1.61ms ± 1% 1.56ms ± 2% -2.59% (p=0.000 n=9+10) Sysbench/SQL/oltp_begin_commit 594µs ± 4% 608µs ± 7% ~ (p=0.089 n=10+10) name old alloc/op new alloc/op delta Sysbench/SQL/oltp_read_only 1.05MB ± 1% 1.04MB ± 0% -0.64% (p=0.004 n=9+10) Sysbench/SQL/oltp_write_only 405kB ± 1% 404kB ± 1% ~ (p=0.720 n=9+10) Sysbench/SQL/oltp_read_write 1.50MB ± 1% 1.51MB ± 1% ~ (p=0.447 n=9+10) Sysbench/SQL/oltp_point_select 35.3kB ± 2% 35.1kB ± 2% ~ (p=0.278 n=9+10) Sysbench/SQL/oltp_begin_commit 18.9kB ± 0% 18.9kB ± 1% ~ (p=0.055 n=10+10) name old allocs/op new allocs/op delta Sysbench/SQL/oltp_read_only 6.96k ± 1% 6.86k ± 1% -1.37% (p=0.000 n=9+10) Sysbench/SQL/oltp_write_only 3.62k ± 1% 3.58k ± 0% -0.96% (p=0.000 n=9+10) Sysbench/SQL/oltp_read_write 11.0k ± 1% 10.9k ± 1% -0.56% (p=0.030 n=9+10) Sysbench/SQL/oltp_point_select 317 ± 1% 312 ± 0% -1.74% (p=0.000 n=9+10) Sysbench/SQL/oltp_begin_commit 140 ± 0% 140 ± 0% ~ (p=1.000 n=10+10) ``` Epic: None Release note: None 134217: sql: fix recently introduced data race r=yuzefovich a=yuzefovich Recently, in a5b8d06 we introduced a possible data race. Namely, that change modifies a proto inside of the physical plan to cap the slice of types during the vectorized operator planning in order to prevent a type schema corruption due to slice aliasing. However, it's been assumed in a couple of places that the physical plan is immutable. This commit clarifies that the physical plan is immutable after its finalization and moves the capping into the finalization. Fixes: #133934. Release note: None Co-authored-by: Xin Hao Zhang <[email protected]> Co-authored-by: David Hartunian <[email protected]> Co-authored-by: Ibrahim Kettaneh <[email protected]> Co-authored-by: Nathan VanBenschoten <[email protected]> Co-authored-by: Yahor Yuzefovich <[email protected]>

blathers-crl · 2024-11-04T20:39:06Z

Based on the specified backports for linked PR #133895, I applied the following new label(s) to this issue: branch-release-24.3. Please adjust the labels as needed to match the branches actually affected by this issue, including adding any known older branches.

_{🦉 Hoot! I am a Blathers, a bot for CockroachDB. My owner is dev-inf.}

The migrations that support the v2 db page runs on finalization. Rather than show a non-functional page in an unfinalized cluster state let's reroute to the legacy page. Epic: CRDB-37558 Fixes: #133843 Release note (ui change): The v2 db pages will only be available post ugprade finalization to 24.3. Prior to that we'll continue to show the legacy page when the cluster is in its unfinalized state.

xinhaoz added the C-enhancement Solution expected to add code/behavior + preserve backward-compat (pg compat issues are exception) label Oct 30, 2024

xinhaoz self-assigned this Oct 30, 2024

xinhaoz added the T-observability label Oct 30, 2024

exalate-issue-sync bot added T-observability and removed T-observability labels Oct 30, 2024

xinhaoz mentioned this issue Oct 30, 2024

ui: reroute v2 db page to legacy page when cluster is unfinalized #133895

Merged

craig bot closed this as completed in f2c93ea Nov 4, 2024

blathers-crl bot added the branch-release-24.3 Used to mark GA and release blockers, technical advisories, and bugs for 24.3 label Nov 4, 2024

blathers-crl bot mentioned this issue Nov 4, 2024

release-24.3: ui: reroute v2 db page to legacy page when cluster is unfinalized #134244

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ui: reroute new db page to legacy page if version error occurs #133843

ui: reroute new db page to legacy page if version error occurs #133843

xinhaoz commented Oct 30, 2024 •

edited by cockroach-jira-scripts

Loading

blathers-crl bot commented Nov 4, 2024

ui: reroute new db page to legacy page if version error occurs #133843

ui: reroute new db page to legacy page if version error occurs #133843

Comments

xinhaoz commented Oct 30, 2024 • edited by cockroach-jira-scripts Loading

blathers-crl bot commented Nov 4, 2024

xinhaoz commented Oct 30, 2024 •

edited by cockroach-jira-scripts

Loading