Releases: FoundationDB/fdb-kubernetes-operator
v1.18.0
foundationdb-ci
FoundationDB CI
v1.18.0
Changes
Operator
- Add FaultDomain as a typed string #1660
- Add new method to get all process group IDs for removal #1659
- Update CronJob and PodDisruptionBudget apiVersion for K8s 1.25 #1658
- Add the current running node for a Pod into the Pod annotations #1657
- Simplify maintenance mode checker logic #1649
- Update maintenance mode information in the update_status controller #1647
- Add the additional processes only once and not multiple times #1643
- Fix e2e docs typo #1642
- Update e2e README.md #1641
- Remove the internal Pod map and use the controller runtime cache #1637
- Make sure we use the force reconcile method if needed and we dump the state if an upgrade is stuck #1636
- Define pr and nightly tests #1635
- Make sure we only set the ResourceTerminating condition if a process group is marked for removal and is fully excluded #1633
- Test upgrade when no remote processes are restarted #1628
- Make sure to only check exclusions if we have 1 active generation #1626
- Test more upgrade scenarios #1613
- Initial support for fault domains in CRD #1608
- Add taint feature to auto replace tainted node #1581
v1.17.0
v1.17.0
Changes
Operator
Make sure we upgrade process groups that are marked for removal during upgrades #1603
Fix the way the resources for storage processes are calculated in the e2e test framework #1609
Test various upgrade scenarios #1580
Add kind setup script for e2e tests #1573
Document localities used and set by the operator #1595
Add support for buggify block removal #1589
Run the exclude command with the highest timeout #1579
Add LogGroup type with specific check #1598
Document the case during an upgrade where the fdbserver is not joining the cluster #1594
Add docs for exclusion handling in the operator #1592
Correct plugin installation steps #1599
Correct statement about setting affinities #1596
Update FDB version used for e2e tests #1578
Fix the way we build the node list in the getNodes function #1582
Update to go 1.19.8 #1577
Add a separate setting for enabling DNS names in pod locality #1301
Add ignore log group in e2e tests #1569
Add an additional wait after installing CRDs in pull request tests #1567
Refactor e2e test suite to make it easier to use the sidecar config #1565
Ensure we only mark the database as unavailable if we can talk to the majority of coordinators #1559
Add test case to validate that the upgrade is stuck when the new binary is missing #1558
Install checksum step for kubectl fdb plugin #1560
Correct the way how the provided images and registry are used #1563
Add an upgrade test case where one sidecar is not ready for the upgrade #1556
Update documentation for e2e tests #1554
Add e2e test setup for operator #1534
adding support for container name in plugin's buggify crashloop command #1515
v1.16.0
foundationdb-ci
FoundationDB CI
v1.16.0
Changes
Operator
Ensure we call exclude only on missing processes #1550
Make sure we always check the sidecar to get the most recent information about the updated ConfigMap #1547
Initial design for replacing Pods running on tainted nodes #1502
Update examples and better document how to run a multi-region FDB cluster #1548
Add kube builder validation checks for a couple of entries. #1544
Adding a way to ignore specific LogGroups in check client compatibility #1526
Fix upgrade logic for multiple storage server per Pod #1538
v1.15.0
foundationdb-ci
FoundationDB CI
v1.15.0
Changes
Operator
- Add script to set up local cluster with kind #1531
- Add validation pattern to ProcessGroupIDPrefix #1529
- Add longer delay after kill command bug Something isn't working #1533
- Ensure we are not blocking if the operator is not able to get the Pod client #1532
- Remove staging phase for version compatible upgrades #1506
- Generate initial cluster file based on "CoordinatorSelection" #1527
- Add Version Check to Redwood Experimental Storage Engine Type bug Something isn't working #1521
- Complement ProcessRole and ProcessClass bug Something isn't working #1522
- Add CRD Validation for CrashLoop #1520
- Fix permission issue for log file writer bug Something isn't working #1524
- Remove the minimum uptime check in the replace incompatible connections subreconciler technical-debt #1517
- Initial round of documenting the upgrade process documentation Improvements or additions to documentation #1511
- fix CVE-2022-41723 dependencies Pull requests that update a dependency file #1516
- Revert fault tolerance check to only check availability #1514
- Fix the internal handling of the lock client for get pending upgrades #1513
- Creating a new type ProcessGroupID #1498
- Correct typo for replaceProcessGroup documentation Improvements or additions to documentation #1510
- Document DoK talk #1509
- Log replacements that the operator would do if the replacement limit would not be blocking it #1505
v1.14.0
foundationdb-ci
FoundationDB CI
v1.13.0
foundationdb-ci
FoundationDB CI
v1.13.0
Changes
Operator
- Make sure we do a delayed reconcile if the incompatible check has an error #1488
- Fix bug in ignore restart logic #1490
- Cdf 17647: patch CVE 2022 41721 #1486
- Improve operator restart logic for upgrades #1474
- Migrate old CrashLoop to new Spec #1484
- Add recommended running env for setup_e2e.sh and fix a broken link in getting_started #1487
- Change crashLoop setup to allow to select container #1472
- Convert some lists in admin_client_mock to sets. #1437
- make: don't target emacs autosave files #1481
v1.12.0
foundationdb-ci
FoundationDB CI
v1.12.0
Changes
Operator
- Remove old Kubernetes version from tests #1471
- Add buggify option to ingore processes during restart #1467
- Add docs for DNS setup documentation Improvements or additions to documentation #1466
- Small clean up #1463
- Add first status check for the analyze command #1461
- Security Updates #1457
- Add update hooks to the mock Kubernetes client #1456
- Ensure that Pods that are stuck in terminating but still have a fdbserver process reporting are marked as failed #1455
- Ensure that the configmap is synced before restarting processes #1454
v1.11.0
v1.10.0
v1.10.0
Changes
Operator
- Add a new reconciler to put the cluster into maintenance mode during rolling upgrade #1275
- Add support for desired and reconciled process groups in status #1365
- Make use of time.duration instead of int for timeout #1371
- Initial changes to support e2e framework #1366
- Remove most of the custom mock implementation #1407
- Make remove incompatible processes more pessimistic #1376
- Upgrade used go version to 1.19.2 #1417
- Add test process count #1362
- Fix nil pointer error #1425
- Refactor Pod models and move strings to constants #1418
- Make use of client library to get the connection string #1421
- Upgrade used go version to 1.19.3 #1427
- Enable error linting in code base #1428
- Refactor admin client #1432
- Initial refactoring of the coordinator selection code #1415
- Allow to define custom create hooks in mock client #1441
- If the cluster is upgraded wait for all processes to be ready #1446
- Remove lock for exclusions #1416
- Add multi-cluster kind setup for testing #1435
- Use deterministic name for cluster file #1374
- Expose mock admin client and lock client #1442r
Docs
v1.9.0
v1.9.0
Changes
Operator
- Reduce API calls to Kubernetes to fetch process group ID information #1295.
- Fix log line #1297.
- Add hot shard tool #1299.
- Update to go 1.18.5 and add docs for local setup with arm64 #1305.
- updating the default time for failure detection [#1312]
- Add better logging for fdb client calls [#1317]
- Add storage engine type [#1318]
- Initial support for server side apply [#1319]
- Add FoundationDB Logo and description [#1321]
- Put correct syntax to get SHA [#1327]
- Correct the way how the connection string is read [#1329]
- Add wait time between actions [#1330]
- Allow downgrades of compatible versions [#1331]
- Update the documentation on per-pod service IPs to reflect where the feature stands [#1332]
- Ignore excluded processes from minimum uptime calculation when doing a rolling bounce [#1333]
- Ignore missing processes during bounce [#1335]
- Only run remove incompatible processes sub reconciler when cluster was upgraded [#1338]
- Add support for recovery state recovered since [#1343]
- Refactor docker images [#1355]
- Add Docker login back for pushing from main [#1356]