-
Notifications
You must be signed in to change notification settings - Fork 3.8k
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
kv/kvserver: TestReplicaStateMachineChangeReplicas failed #69229
Labels
branch-master
Failures and bugs on the master branch.
C-test-failure
Broken test (automatically or manually discovered).
O-robot
Originated from a bot.
Comments
cockroach-teamcity
added
branch-master
Failures and bugs on the master branch.
C-test-failure
Broken test (automatically or manually discovered).
O-robot
Originated from a bot.
labels
Aug 23, 2021
craig bot
pushed a commit
that referenced
this issue
Sep 2, 2021
69720: build: increase pebble nightly metamorphic test runtime r=nicktrav a=nicktrav Increase the runtime of the Pebble nightly metamorphic test run from 1h to 3h. More test time increases the chance of catching a test failure. See #69414. Release justification: non-production code change Release note: None 69730: kv: deflake TestReplicaStateMachineChangeReplicas r=nvanbenschoten a=nvanbenschoten Fixes #69229. The test was reaching below Raft and adding a peer on a node that does not exist. With a large enough pause between `r.raftMu.Unlock()` and `testContext` teardown, it was possible for the replica to be ticked, which would cause it to attempt to reach out and heartbeat this node. This could cause issues. In fact, the `testContext` uses a dummy `RaftTransport`, so any attempt to send traffic to another Raft peer would panic. This commit resolves the issue by giving testContext a slightly more real version of a RaftTransport (i.e. one configured with a nodedialer). This feels a bit wrong, as a testContext should really never need to use a RaftTransport, but given how little impact this has on the dependency structure of a testContext and given that the testContext was already doing something similar with gossip, I think this is fine. Release justification: test fix. 69780: ui: Combining Type for Aggregated Node Status r=nathanstilwell a=nathanstilwell This change is to alleviate an uncaught error thown by when applying an aggregated status to a group of nodes. When nodes are aggregated into rows a summary status is applied to a row. To apply the appropriate style to the badge that represents the status, the function getBadgeTypeByNodeStatus is called. This function was being called with a status of type AggregatedNodeStatus rather than LivenessStatus causing the error. To fix this I combined the two functions getBadgeTypeByNodeStatus and getBadgeTypeByNodeStatusAggregated into a single function that compares the status parameter to the two enum types to determine status. Instead of throwing an error for the default case, I am simply returning default for the badge type. While this may cause a visual bug under the right circumstances, at least it won't crash the app. While I was in here, I also added more TypeScript annontations to remove warnings. fixes #69767 69788: server: add NodeIds to DatabaseDetails and TableStats responses r=xinhaoz a=xinhaoz Related to: #63391 Previously, DatabaseDetails and TableStats did not return a list of nodes that store their data. This commit adds an ordered list of node ids to each of the responses. This is necessary to report node/region information in the DB console databases pages. For the database details endpoint, we return the list of nodes only if stats are requested. Release justification: low risk, high benefit changes to existing functionality Release note (api change): A list of node ids representing the nodes that store data for the database has been added to the stats field in the databases details endpoint under `nodeIds`. Database details must be requested with include_stats set to true, e.g. `/_admin/v1/databases/{database}?include_stats=true` Similarly, `nodeIds` has also been added to the table stats endpoint, which is an ordered list of node ids which stores the table data: `/_admin/v1/databases/{database}/tables/{table}/stats` 69800: logictestccl: restore the flake fix r=yuzefovich a=yuzefovich Restore the fix from fa4e16d which was just lost in b86022e. Addresses: #68395 (comment). Release note: None Release justification: testing only change. 69802: vendor: bump Pebble to 1f862845897e r=bananabrick a=jbowens ``` 1f86284 vfs/atomicfs: add Marker type 8731fd6 sstable: Fix bug in dynamic readahead 085aaac sstable: support block checksum validation for entire SSTables 6236eba pebble: Unflake TestCacheEvict 38dd75e ci: remove support for Go v1.14; remove TravisCI ``` Release note: none Release justification: non-production code changes, low-risk high benefit to existing functionality (8731fd6) Co-authored-by: Nick Travers <[email protected]> Co-authored-by: Nathan VanBenschoten <[email protected]> Co-authored-by: Nathan Stilwell <[email protected]> Co-authored-by: Xin Hao Zhang <[email protected]> Co-authored-by: Yahor Yuzefovich <[email protected]> Co-authored-by: Jackson Owens <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Labels
branch-master
Failures and bugs on the master branch.
C-test-failure
Broken test (automatically or manually discovered).
O-robot
Originated from a bot.
kv/kvserver.TestReplicaStateMachineChangeReplicas failed with artifacts on master @ 61bd543ba7288c8f0eed6cddded7b219c9d1fcd4:
Fatal error:
Stack:
Log preceding fatal error
Reproduce
To reproduce, try:
Parameters in this failure:
This test on roachdash | Improve this report!
The text was updated successfully, but these errors were encountered: