Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Browse files
Browse the repository at this point in the history
…76591 75809: [CRDB-12226] server, ui: display circuit breakers in problem ranges and range status r=Santamaura a=Santamaura This PR adds changes to the reports/problemranges and reports/range pages. Ranges with replicas that have a circuit breaker will show up as problem ranges and the circuit breaker error will show up as a row on the status page. Release note (ui change): display circuit breakers in problems ranges and range status Problem Ranges page:  Range status page:  76278: Add cluster version as feature gate for block properties r=dt,erikgrinaker,jbowens a=nicktrav Two commits here - the first adds a new cluster version, the second makes use of the cluster version as a feature gate (and update various call sites all over the place). --- pkg/clusterversion: add new version as feature gate for block properties Prior to this change, the block properties SSTable-level feature was enabled in a single cluster version. This introduced a subtle race in that for the period in which each node is being updated to `PebbleFormatBlockPropertyCollector`, there is a brief period where not all nodes are at the same cluster version, and thus store versions. If nodes at the newer version write SSTables that make use of block properties, and these tables are consumed by nodes that are yet to be updated, the older nodes could panic when attempting to read the tables with a format they do not yet understand. While this race is academic, given that a) there are now subsequent cluster versions that act as barriers during the migration, and b) block properties were disabled in 1a8fb57, this patch addresses the race by adding a second cluster version. `EnablePebbleFormatVersionBlockProperties` acts as a barrier and a feature gate. A guarantee of the migration framework is that any node at this newer version is part of a cluster that has already run the necessary migrations for the older version, and thus ratcheted the format major version in the store, and thus enabled the block properties feature, across all nodes. Add additional documentation in `pebble.go` that details how to make use of the two-version pattern for future table-level version changes. --- pkg/storage: make MakeIngestionWriterOptions version aware With the introduction of block properties, and soon, range keys, which introduce backward-incompatible changes at the SSTable level, all nodes in a cluster must all have a sufficient store version in order to avoid runtime incompatibilities. Update `storage.MakeIngestionWriterOptions` to add a `context.Context` and `cluster.Settings` as parameters, which allows for determining whether a given cluster version is active (via `(clusterversion.Handle).IsActive()`). This allows gating the enabling / disabling of block properties (and soon, range keys), on all nodes being at a sufficient cluster version. Update various call-sites to pass in the `context.Context` and `cluster.Settings`. --- 76348: ui: downsample SQL transaction metrics using MAX r=maryliag a=dhartunian Previously, we were using the default downsampling behavior of the timeseries query engine for "Open SQL Transactions" and "Active SQL Statements" on the metrics page in DB console. This led to confusion when zooming in on transaction spikes since the spike would get larger as the zoom got tighter. This PR changes the aggregation function to use MAX to prevent this confusion. Resolves: #71827 Release note (ui change): Open SQL Transactions and Active SQL Transactions are downsampled using MAX instead of AVG and will more accurately reflect narrow spikes in transaction counts when looking and downsampled data. 76414: spanconfig: teach the KVAccessor about system span configurations r=arulajmani a=arulajmani First 3 commits are from #76219, this one's quite small -- mostly just tests. ---- This patch teaches the KVAccessor to update and get system span configurations. Release note: None 76538: ui: Use liveness info to populate decommissioned node lists r=zachlite a=zachlite Previously, the decommissioned node lists considered node status entries to determine decommissioning and decommissioned status. This changed in #56529, resulting in empty lists. Now, the node's liveness entry is considered and these lists are correctly populated. Release note (bug fix): The list of recently decommissioned nodes and the historical list of decommissioned nodes correctly display decommissioned nodes. 76544: builtins: add rehome_row to DistSQLBlocklist r=mgartner,otan a=rafiss fixes #76153 This builtin always needs to run on the gateway node. Release note: None 76546: build: display pebble git SHA in GitHub messages r=rickystewart a=nicktrav Use the short from of the Git SHA from the go.mod-style version in DEPS.bzl as the Pebble commit. This ensures that failure messages created by Team City link to a GitHub page that renders correctly. Release note: None 76550: gen/genbzl: general improvements r=ajwerner a=ajwerner This change does a few things: * It reworks the queries in terms of eachother in-memory. This is better than the previous iteration whereby it'd generate the results and then rely on the output of that query. Instead, we just build up bigger query expressions and pass them to bazel using the --query_file flag. * It avoids exploring the pkg/ui directory (and the pkg/gen directory) because those can cause problems. The pkg/ui directory ends up bringing in npm, which hurts. * It stops rewriting the files before executing the queries. It no longer needs to rewrite them up front because they aren't referenced by later queries. * It removes the excluded target which was problematic because those files weren't properly visible. Fixes #76521 Fixes #76503 Release note: None 76562: ccl/sqlproxyccl: update PeekMsg to return message size instead of body size r=JeffSwenson a=jaylim-crl Informs #76000. Follow-up to #76006. Previously, PeekMsg was returning the body size (excluding header size), which is a bit awkward from an API point of view because most callers of PeekMsg immediately adds the header size to the returned size previously. This commit cleans the API design up by making PeekMsg return the message size instead, i.e. header inclusive. At the same time, returning the message size makes it consistent with the ReadMsg API since that returns the entire message. Release note: None 76591: bazel: update shebang line in `sql-gen.sh` r=rail a=rickystewart Release note: None Co-authored-by: Santamaura <[email protected]> Co-authored-by: Nick Travers <[email protected]> Co-authored-by: David Hartunian <[email protected]> Co-authored-by: arulajmani <[email protected]> Co-authored-by: Zach Lite <[email protected]> Co-authored-by: Rafi Shamim <[email protected]> Co-authored-by: Andrew Werner <[email protected]> Co-authored-by: Jay <[email protected]> Co-authored-by: Ricky Stewart <[email protected]>
- Loading branch information
