v1.18.0

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

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

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

v1.14.0

Changes

Operator

• Improve the way how the operator handles partitial upgrades #1491
• extending the cordon command for all clusters #1493
• Update dependencies #1499

v1.13.0

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

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

v1.11.0

Changes

Operator

• Expose Pod Mock Client for testing #1451
• Initial changes to remove update Pod config in bounce reconciler #1372
• If configure database is set to false skip all work #1433

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

• Fix broken link in sidecar README.md #1364
• Update automatic_replacements.md #1443

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 [FoundationDB#1312]
• Add better logging for fdb client calls [FoundationDB#1317]
• Add storage engine type [FoundationDB#1318]
• Initial support for server side apply [FoundationDB#1319]
• Add FoundationDB Logo and description [FoundationDB#1321]
• Put correct syntax to get SHA [FoundationDB#1327]
• Correct the way how the connection string is read [FoundationDB#1329]
• Add wait time between actions [FoundationDB#1330]
• Allow downgrades of compatible versions [FoundationDB#1331]
• Update the documentation on per-pod service IPs to reflect where the feature stands [FoundationDB#1332]
• Ignore excluded processes from minimum uptime calculation when doing a rolling bounce [FoundationDB#1333]
• Ignore missing processes during bounce [FoundationDB#1335]
• Only run remove incompatible processes sub reconciler when cluster was upgraded [FoundationDB#1338]
• Add support for recovery state recovered since [FoundationDB#1343]
• Refactor docker images [FoundationDB#1355]
• Add Docker login back for pushing from main [FoundationDB#1356]