Repository navigation
Releases
kividb-operator v0.7.1
A patch release of 0.7.0. The default engine is kividb 1.2.1. Upgrade
now if you run engine 1.2.1, or will: since 1.2.1 the engine refuses a
write pause it cannot take at once, and operator 0.7.0 then switched over
without the pause, so a write the old master acknowledged could be lost
(#81). 0.7.1 also fences a force-deleted master's last writes before a
failover (#53), keeps a restore from mixing the snapshot with the earlier
cluster's append-only file (#58), fixes the images' CVEs (#76), adds an
OLM bundle for OperatorHub.io, and fixes three install paths.
Read before upgrading: every cluster restarts, and a master with no
replica restarts in place. The default agent image moves to 0.7.1, and
the default engine to 1.2.1, so every KividbCluster that leaves
spec.agentImage or spec.image unset gets a rolling restart when the
operator is upgraded. A master with replicas hands its role to an in-sync
replica first, as in every rollout. A master with none is deleted and
restarted on its volume, with no handover:
replicas: 0(one pod): the cluster serves nothing until the pod is
back.- A sharded cluster with
replicas: 0: each shard's master restarts
in turn. Its slots are unavailable while it does, and if it is down for
longer than the node timeout (5 s by default) withrequireFullCoverage
on (the default), the whole cluster refuses commands until it is back. - Data: with the append-only file on (the default since 0.7.0) the
master reloads everything it acknowledged. Withappendonly: "no"it
comes back with its shutdown snapshot only, and with nothing saved since
its last snapshot if that save does not finish within
terminationGracePeriodSeconds.
Before upgrading, give such clusters or shards at least one replica and
let it sync, or accept the restart: with the append-only file on, or with
the shutdown snapshot only. (Turning the append-only file on is a config
change, so it restarts such a master in place too.) To put the restart off, pin
both spec.image: docker.io/kividb/kividb:v1.2.0 and
spec.agentImage: docker.io/kividb/kividb-operator-agent:0.7.0 first:
nothing else in the pod template changes in 0.7.1. The 0.7.0 agent works
with the 0.7.1 operator, without the agent's fixes: it cuts every pause at
5 s, which aborts a switchover that needs longer (#81), and its restore
lacks the #58 fix. Unpin once the cluster has a replica or its
append-only file.
Engine 1.2.1 has upgrade notes of its own. The rollout restarts
replicas before their master, as 1.2.1 requires (a 1.2.0 replica cannot
full-sync from a 1.2.1 master that holds a key 1.2.0 cannot read). Read
the engine's "Before you upgrade"
first: among others, ACL categories are Redis's from the moment a pod
runs 1.2.1, GEORADIUS is refused on replicas, and there is no going
back to 1.2.0 once 1.2.1 has saved such a key. Do not change
spec.shards during the rollout: a slot move from a 1.2.0 node loses
the moved time series' compaction rules.
The CRD change is additive, and upgrading from 0.7.0 is safe. The
kividbclusters status gains pods[].podIP and fencedMaster (name,
podIP, runID), and new conditions (Switchover); nothing is removed
or renamed, and the other CRDs do not change. Apply the new
kividbclusters CRD before helm upgrade: Helm does not upgrade it.
Until it is applied, the API server drops the new fields and failovers do
not hold the old master's writes (FailoverUnfenced Warning). The CRDs
are attached to the GitHub Release. Coming from 0.6.x, read 0.7.0's
"Read before upgrading" first.
Security
- The images had golang.org/x/net v0.60.0 (CVE-2026-78661, high;
CVE-2026-97045, medium) and an out-of-date Debian 12 base (#76).
golang.org/x/net is now v0.62.0, the other golang.org/x modules are
current, the binaries are built with Go 1.26.11, and the images run on
gcr.io/distroless/static-debian13:nonroot(still no shell, still UID
65532). CI and the release workflow scan every image with Trivy and fail
on a fixable HIGH or CRITICAL.
Changed
- Default engine
docker.io/kividb/kividb:v1.2.1(it wasv1.2.0):
in the controller, the CRD, the chart, the GUI, the samples and the docs.
See "Read before upgrading" above. - Default agent image
docker.io/kividb/kividb-operator-agent:0.7.1.
A cluster that leavesspec.agentImageunset rolls to it in the same
rollout. The switchover and restore fixes below need it. - The e2e suite tests kividb 1.2.1 by default.
Added
- An OLM bundle for OperatorHub.io.
make bundle VERSION=… IMG=…builds
the bundle fromconfig/manifests(operator-sdk v1.36.0, kustomize v5.4.3),
andmake scorecardruns the scorecard tests. The bundle validates for
Kubernetes 1.27, 1.32 and 1.33 and installs through OLM 0.28.0 in
AllNamespaces mode.
Fixed
-
The plain-YAML install and
kubectl apply -k config/defaultmade the
manager Deployment with different selectors (#86). Applying one over the
other failed withspec.selector: ... field is immutable.
config/manager/manager.yamlnow selects onapp.kubernetes.io/managed-by
too, asconfig/defaultdoes; whatconfig/defaultand the OLM bundle
render is unchanged. A Deployment made before this fromconfig/manager
alone, by the plain-YAML install orkubectl apply -k config/manager, has
the old selector: delete it once (kubectl delete deploy -n kividb-operator-system kividb-operator-manager) before applying the new
manifests.
(#88, by SaiPisey2) -
docs/INSTALL.md's install without kustomize exited 1 (#67).
kubectl apply -f config/rbac/applied the RBAC and then failed on
config/rbac/kustomization.yaml, and on a new cluster the namespace did
not exist. The steps now create the namespace and name the RBAC files.
(#85, by SaiPisey2) -
A switchover no longer hands over without holding writes when the pause
could not take hold (#81). Since kividb 1.2.1 (kividbio/kividb#629),
CLIENT PAUSE ... WRITEwaits up to 5 s for the writes already in progress
and refuses, pausing nothing, if one has not finished (a bigFLUSHALL, a
DELof a huge key, a long script). The switchover took that refusal, and
a pause that got no answer within the agent's 5 s, for an engine before
kividb 1.1: it promoted the target without waiting for it to catch up,
and raisedSwitchoverWithoutPausetelling the user to upgrade an engine
that was already new. A write the old master acknowledged then could be
lost. Now such a switchover is called off (SwitchoverAborted) and tried
again; a pause that got no answer is lifted, in case it landed. Only an
engine that answers that it does not know the command is switched over
without the pause: every kividb has it, so any other refusal (LOADING,
BUSY) aborts too. The pause call is given 7 s, past the engine's wait
(or--agent-timeout, if longer), and the agent is asked to wait 6 s:
its exec request takes atimeoutMs(capped at 15 s), which the
operator sends for the pause only. An agent before this release ignores
it and cuts the pause at 5 s, which now aborts the switchover rather than
dropping the pause. The engine counts the pause from the request, so it
now lasts the call, the catch-up and a margin (17 s), and the old master
is a replica before it runs out. Aborted switchovers back off: 30 s,
doubling with each abort in a row up to 10 min, and 2 s between the tries
for a master pod being deleted; aSwitchovercondition (False,
Aborted) says why and when the next try is, until one goes through.
The failover fence (#53) waits for a pause the same way once the old
master has answered, and tries again after a refusal: every try is given
the whole call, and new ones start for 3 s after the first, so the fence
takes at most two pause calls and that window against a live master that
cannot pause (17 s with the default--agent-timeout, more with a longer
one). A pause that got no answer may have landed, so that
master is recorded instatus.fencedMasterand demoted after the
promotion like a fenced one. TheSwitchovercondition and the count of
aborts go away once no switchover is wanted any more, or the master is
another pod.
(#82, by lofoneh) -
kubectl apply -k config/defaultinstalled noKividbDbOpsCRD.
config/crd/kustomization.yamldid not list it, so the documented install
could not runKividbDbOpsoperations. After applying it again, restart
the manager (kubectl rollout restart): it looks for the CRD once, at
start-up. -
The full
KividbConfigsample was rejected by its own CRD. It set
tls: {enabled: false}, and the CRD requirescertSecretRefwhenevertls
is present. The sample now leavestlsout. -
A restore over a kept volume could mix the snapshot with the earlier
cluster's append-only file (#58). The restore Job removes the earlier
cluster'sdump.kdb.gen, and without it engine 1.2.0 and 1.2.1 both
replayappendonly.aofoverdump.kdb. The Job removed the earlier
cluster'sappendonly.aofonly after it had renamed the restored
dump.kdbin, so a Job stopped between the two (killed, evicted, or a
removal that failed) left the restored dump with the earlier cluster's
log. A cluster started on that volume, oncespec.bootstrapFromSnapshot
was removed after a failed bootstrap, came up with the writes of both.
Every file of the earlier cluster's engine now goes before a restored
file takes its name:appendonly.aofand everyappendonly.aof.*,
everydump.kdb.*(its state and temp files),dump.rdb, and the
superseded.*logs that engine 1.2.1 keeps aside. The engine never
removes those: they held the volume's space, and every start named...
kividb-operator v0.7.0
Sharded clusters (spec.shards). The default engine is kividb 1.2.0.
Read before upgrading: the agent's port (8081) is no longer on the
<cluster>-master and <cluster>-replicas Services, and the agent's API
needs a token. Anything that called an agent through those Services, other
than the operator's own backup Jobs, has to change. The pods roll once.
New backups go to <pathPrefix>/<namespace>.<cluster>/ instead of
<pathPrefix>/<cluster>/: widen any IAM or bucket policy that limits the
backup credentials to the old folder, and adjust scripts and lifecycle
rules that match the old keys (see Fixed).
Apply the new CRDs before helm upgrade: Helm does not upgrade them,
and the KividbCluster, KividbConfig, KividbDbOps and KividbSnapshot
CRDs changed. Until they are applied, the API server drops the new fields
(spec.shards among them) and the checks that read them never fire. The
CRDs are attached to the GitHub Release.
Read before upgrading: 0.7.0 targets kividb engine 1.2.0.
- A cluster without
spec.imagegets an engine upgrade to the new
default image,docker.io/kividb/kividb:v1.2.0, as a rolling restart:
replicas first, masters last, each master only after an in-sync replica
has taken over. Pinspec.imageto stay on 1.1.x for now. - No going back to 1.1.x once a 1.2 pod has saved a snapshot that
holds a time series with compaction rules: 1.1.x cannot read it. Take
a backup before the upgrade if you may need to downgrade. - During the rollout, 1.1.x replicas of a 1.2 master log errors. A 1.2
master sends a transaction, and a script that writes more than once, to
its replicas betweenMULTIandEXEC; a 1.1.x replica still applies
every write, but logsReplicated command error (non-fatal): ERR unknown command 'MULTI'(and'EXEC') for each, until it is upgraded too. - The append-only file is on by default (
appendonly yes,
appendfsync everysec), for every cluster whoseKividbConfigdoes not
setappendonlyitself. It comes with the rolling restart above. Writes
go through the file, so expect fewer writes a second at high
concurrency, and size the volumes for the log too.appendonly: "no"in
aKividbConfigkeeps the old behaviour, with aPersistencecondition
and a Warning Event that say what it risks: without it, a killed engine
comes back with only its last shutdown or full-sync snapshot, often
nothing (see Fixed). - A
KividbConfigdirective the engine refuses stops the server. 1.2
accepts Redis's own settings with a warning and applies some it used to
ignore, but a name that is neither KiviDB's nor Redis's, or a bad value,
is still fatal. A rolling change stalls safely after the first pod; a new
cluster or a cold restart has every pod down. See
CONFIGURATION.md. - A
KividbSnapshotConfigwhosespec.schedulehas aTZ=or
CRON_TZ=prefix is now refused asInvalidSpec(#47): a CronJob
does not accept one. Every cluster that references it holds back its
apply steps until it is fixed, though failover still runs. Remove the
prefix before upgrading. - Reshard only under light write load. Engine 1.2.0 can lose writes
while slots move between shards under heavy writes
(kividbio/kividb#331). That covers changingspec.shards(scaling in
or out, which usesCLUSTER MIGRATION IMPORT) andredis-cli --cluster reshard. It was seen only under heavy CPU contention. Until engine
1.2.1, changespec.shardswhile writes are light, or pause writes.
Engine 1.2.1 fixes it: setspec.imageto it once it is released.
Security
- The agent's API needs a token. In 0.6.0 any client that could reach
port 8081 of a kividb pod could promote it, repoint its replication, or
run any command through/exec; and the port was on the master and
replica Services, which can be LoadBalancers. Now every endpoint but
/healthz,/readyzand/metricsrequiresAuthorization: Bearer <token>, and an agent without a token refuses them. The operator keeps
the token in a Secret in its namespace (created with a random value at
its first start;--agent-token-secret, the chart's
<release>-agent-token) and copies it into each cluster's auth Secret.
The operator, the backup Jobs and the GUI send it; the GUI reads it from
the Secret mounted as a file. A new token rolls the pods, as a new
password does. - The agent port is on the headless Service only, not on the master
and replica Services.
Added
spec.shards: sharded clusters in kividb's cluster mode, one
StatefulSet per shard (<cluster>-<shard>), a PodDisruptionBudget per
shard, and ClusterIP Services only: one per pod, one per shard,
<cluster>-replicas,<cluster>for the whole cluster, and a headless
Service per shard.spec.clustersets the node timeout, full coverage,
min-replicas-to-writeand the bus secret (generated when unset). The
operator owns every cluster-mode directive. A cluster is sharded or not
from its creation; up to 24 nodes for now.- Sharded clusters form and look after themselves. The operator
introduces the nodes (CLUSTER MEET), assigns the slots evenly across the
shards once every pod is up, puts each shard's replicas behind its
master, and has every node announce its pod's DNS name. Failover is the
engine's: the operator reports it and moves the role labels, and never
promotes in cluster mode. It re-introduces nodes after IP changes (a
cold restart), and forgets the old ID of a pod that rejoined with a new
one, never a node that owns slots. The agent gains/cluster/status,
/meet,/addslots,/replicate,/forgetand/announce, all behind
the agent token. - Rolling restarts of sharded clusters, for a template change or a
KividbDbOpsrestart: one pod at a time across the cluster, replicas
first, and each shard's master only after its most caught-up replica
has taken over withCLUSTER FAILOVER. The agent has a
/cluster/failoverendpoint, and reports whether a replica's link to
its master is up;status.pods[].syncedis set for sharded clusters. - Changing a sharded cluster's shard count. Edit
spec.shardsand
the operator adds shards, or drains and removes them, moving only the
slots it has to: one move at a time, each an atomicCLUSTER MIGRATION IMPORTon the destination master (noASKfor clients). Phase
Rebalancing, with the plan, the move under way and the last error in
status.cluster.rebalance.status.cluster.shardsis the count the
slots are spread over. Removed shards' PVCs are deleted once their
slots have all moved out. The agent has
/cluster/migration/importand/cluster/migration/status. - Backups of sharded clusters. A run backs up every shard at once,
each from a pod with the configured role, under
<namespace>.<cluster>/<run>/shard-<i>-<pod>.tar.gz. Amanifest.json
names each shard's object, source, master node ID and slots, and is
written only when every shard succeeded. Retention counts complete runs
and prunes whole runs.KividbSnapshot.status.shards[]lists the
shards, andobjectKeyis the manifest. The agent has
/backup/manifest, and/backuptakes a run and shard. Restoring a
sharded cluster is not supported yet. - Access to a sharded cluster from outside Kubernetes. Expose each pod
with a Service of your own (a LoadBalancer, a NodePort) labelled
kividb.io/announce: "true", and the pod's node announces that
Service's address (its load-balancer address, or the
kividb.io/announce-addressannotation) instead of its DNS name, so
cluster clients outside followMOVEDto an address they can reach.
The operator only reads these Services. TheExternallyAccessible
condition says whether every pod is exposed (PartiallyExposedwhen
only some are) and what is wrong with a Service that cannot be used. - The console shows sharded clusters.
- Shards: one card per shard, with its slots, master and replicas, and each replica's lag against its own shard's master.
- Slot map: the slots coloured by shard; the ones moving are striped.
- Rebalance: progress while the shard count changes.
- Tiles: slots and
cluster_state. - Actions: +/− Shard. Promote runs
CLUSTER FAILOVERon the replica. Take snapshot starts a backup run of every shard, as a Job from the backup CronJob, so the GUI's ClusterRole gainscreateonjobs.
spec.cluster.replicaValidityFactor:cluster-replica-validity-factor,
which engine 1.2 enforces: a replica whose link to its master has been
down for longer thannodeTimeoutMs × replicaValidityFactordoes not
stand for election. Unset, the operator writes nothing and the engine's
default, 10, applies (with the default 5 s node timeout, a replica cut
off for more than about a minute does not take over);0lets every
replica stand. Needs engine 1.2: older engines do not start with the
directive. AKividbConfigmay not set it.- Handovers that do not happen are reported. Engine 1.2 pauses a
shard master's writes for up to 10 s duringCLUSTER FAILOVER, and the
replica gives the failover up after 5 s (Manual failover abandoned).
TheShardSwitchoverEvent now marks only the first request of a
handover; aWarningShardSwitchoverAbandonedfollows when the master
has not moved 15 s after a request. The operator asks at most 5 times
in a row (then again after 10 minutes), and theShardSwitchover
condition shows the shard, the pods and the attempt while a handover is
pending (InProgress,Retrying,Stalled). - A
KividbDbOpsrestart no longer waits forever for a master to hand
over. After 5 minutes it fails, says where to look, and withdraws its
step-down request.status.restart.handoverStartTimerecords when it
st...
kividb-operator v0.6.1
A patch release of 0.6.0. The default engine is kividb 1.1.3, a
security release of kividb 1.1.
Security
- Default engine
docker.io/kividb/kividb:v1.1.3(it wasv1.1.2):
in the controller, the CRD, the chart, the GUI, the samples and the docs.
AKividbClusterthat leavesspec.imageunset moves to 1.1.3 when the
operator is upgraded, the way every rollout goes: the replicas are
restarted first, then the operator switches over to an up-to-date replica,
and the old master is restarted last. Setspec.imageto stay on a
version.
Changed
- Default agent image
docker.io/kividb/kividb-operator-agent:0.6.1.
A cluster that leavesspec.agentImageunset rolls to it in the same
rollout. - The e2e suite tests kividb 1.1.3 by default.
kividb-operator v0.6.0
A new management GUI. The default engine stays kividb 1.1.2.
Read before upgrading: with GUI authentication on, the GUI now has its
own sign-in page and session cookie in place of the browser's Basic-auth
box. Scripts that call the API with Basic auth (curl -u) keep working.
Added
- A sign-in page (
/login). Signing in sets a session cookie
(HttpOnly, SameSite=Lax, Secure over HTTPS, 12 hours), signed with a
random key made when the GUI starts: nothing in it is derived from the
password, and a restart (which changing the password needs) ends every
session. Pages redirect to it when signed out,
and come back to where you were; the API answers 401 without the header
that makes browsers show their own login box. Ten failed attempts from
one address lock it out for five minutes. Sign out is in the top
bar. Basic auth still works for scripts. - Dashboard overview. Summary tiles (clusters; healthy, needing
attention, in error; memory, commands per second and clients across all
pods), a search box and a health filter, and for each cluster its engine
version and build, a pod bar, and sparklines of the last hour's commands
per second and memory. The figures come from the GUI's own scrapes, so
the dashboard is still oneListofKividbClusters. - Topology on a cluster's page: the master and its replicas, each with
its readiness, memory, keys, replication offset and node, and how far each
replica is behind the master. - Charts with axes, a hover readout of every pod, a legend, and a 1 h /
6 h / 24 h range: commands per second, memory and clients per pod, and
each replica's replication lag (which replaces the raw offset chart). - New cluster (
/new): a form for the name, namespace, build (standard,
lua, tls, full), replicas, storage, memory and CPU, storage class, master
service type, backups and monitoring. The manifest beside it is the API
server's dry run of the form, updated as you type, so a mistake shows
before anything is created; it can be copied forkubectl apply. Picking
a build sets the matching image as well asspec.variant(the operator
never derives one from the other). Creating needs GUI authentication, as
every other change does.POST /api/clusters(?dryRun=1for the
preview) andGET /api/create-optionsback it. - Light and dark themes, following the system setting until you pick
one; a layout for phones and tablets. - Logs: a line count, follow mode, highlighting, and errors and warnings
in colour. Console (the RESP explorer): replies printed as
redis-cliprints them, quoted arguments, and history on ↑ / ↓. GET /api/session: whether the GUI accepts changes. On a read-only GUI
every button that would only be refused is disabled, and says why.
Changed
- Confirmations are in-page dialogs, not the browser's
confirmand
prompt; deleting a cluster asks for its name. Results show as toasts. - The cluster page's tabs are Overview, Metrics, Logs, Console, Backups,
and Events & operations. - The GUI's ClusterRole (chart and
config/gui/rbac.yaml) addscreateon
kividbclustersandget/listonstorageclasses, for the new-cluster
form. - Default agent image
docker.io/kividb/kividb-operator-agent:0.6.0.
AKividbClusterthat leavesspec.agentImageunset rolls its pods to
it, one by one, when the operator is upgraded. - The chart is signed with cosign, and Artifact Hub's repository metadata
is published with it (from #8).
kividb-operator v0.5.2
A security release. The default engine stays kividb 1.1.2.
Security
github.com/klauspost/compressv1.17.9 -> v1.18.7
(CVE-2026-63209, GHSA-259r-337f-4rfw: an integer overflow in
s2.NewDict). It reaches the agent throughminio-go, which the agent
uses for S3 backups. govulncheck finds no call to the vulnerable code.golang.org/x/netv0.56.0 -> v0.60.0 andgolang.org/x/crypto
v0.53.0 -> v0.57.0, which fix the advisories govulncheck reports for
them (none of the vulnerable code is called either), withx/sys,
x/termandx/textmoved along.
Changed
- Go 1.26.
x/netv0.60.0 andx/cryptov0.57.0 need it, sogo.mod
now saysgo 1.26.0and CI builds with 1.26, the same version as the
release images (golang:1.26-bookworm). Building the operator from
source needs Go 1.26. - Default agent image
docker.io/kividb/kividb-operator-agent:0.5.2.
AKividbClusterthat leavesspec.agentImageunset rolls its pods to
it, one by one, when the operator is upgraded.
kividb-operator v0.5.1
The default engine is kividb 1.1.2, a patch release of kividb 1.1.0
that fixes a RESTORE that could crash the server and expired keys piling
up under a steady load of writes with a TTL (kividb 1.1.1 was withdrawn for
the second). Validated on Azure Kubernetes Service (three nodes) with this
operator's code against kividb 1.1.2's: compat variants (standard, TLS,
Lua, full), a master killed under write load, snapshot chaos, monitoring,
RESP/ACL/SELECT, function libraries through a replica restart, a failover
and a restore, backup to S3 and restore, a sustained load of
SET ... EX 5 on a master and two replicas, and in-place upgrades from
kividb 1.1.0 and 1.1.1 to 1.1.2 under write load (no password, ACL users,
TLS). No acknowledged write was lost in any of them.
Changed
- Default engine
docker.io/kividb/kividb:v1.1.2(it wasv1.1.0):
in the controller, the CRD, the chart, the GUI, the samples and the docs.
AKividbClusterthat leavesspec.imageunset is upgraded to 1.1.2
when the operator is upgraded: the operator resolves the default on
every reconcile, so the pods roll one by one as for any image change. Set
spec.imageto stay on a version. kividb's image is distroless since
1.1.2 (no shell, noredis-cliin the container); the operator only
talks to kividb over RESP, through its agent, so nothing changes for it. - Default agent image
docker.io/kividb/kividb-operator-agent:0.5.1.
A cluster that leavesspec.agentImageunset rolls to it too. - The e2e suite tests kividb 1.1.2 by default.
Fixed
kividb-operator v0.5.0
Validated on Azure Kubernetes Service (three nodes) against kividb v1.1.0:
a cluster without a password, a password-protected cluster with ACL users
upgraded in place from v1.0.5 to v1.1.0 under write load, a master killed
under write load, and the TLS build with a client CA. No acknowledged write
was lost in any of them.
Changed
- Default engine
docker.io/kividb/kividb:v1.1.0. Images now come
from Docker Hub (docker.io/kividb/...) and GHCR (ghcr.io/kividbio/...),
with the same tags in both; nothing is published to Quay any more. The
operator, agent and GUI images, the chart, the CRD defaults, the samples
and the docs point at Docker Hub. The Helm chart is at
oci://registry-1.docker.io/kividb/kividb-operator-chartand
oci://ghcr.io/kividbio/kividb-operator-chart. - Default agent image
docker.io/kividb/kividb-operator-agent:0.5.0. - The release attaches the CRDs (
config/crd/bases/*.yaml) as well as the
packaged chart, asdocs/RELEASING.mdalready told users to apply them
from there.
Fixed
- A cluster without a password served nobody on kividb v1.1.0.
kividb 1.1.0 enforces protected mode: with no password on thedefault
user and nobind, it refuses every client that is not on the pod's
loopback, the other pods' replication included. The agent, on loopback,
still reported the pods healthy, so such a cluster showedRunningwhile
every client and replica gotDENIED. The operator now sets
bind 0.0.0.0, the address kividb always listened on, unless you
configurebindorprotected-modeyourself. Existing clusters are
rolled onto the new configuration by the operator upgrade. - The chart's
agent.imagewas not used. Every cluster without
spec.agentImagepulled the image built into the operator, so mirroring
the images to a private registry through the chart values left the
sidecars unpullable. The chart now passesagent.imageto the manager's
new--default-agent-imageflag. - The TLS port was on no Service. The pods listened on it, but the
master, replica and headless Services carried only the kividb and agent
ports, so TLS clients could reach kividb only by pod IP. With TLS on, all
three now carry it, namedtls.
Upgrade notes
- kividb v1.1.0 with a TLS CA requires client certificates. With
tls.certSecretRef.caKeyset, a TLS client without a certificate signed
by that CA is refused; earlier engines let it in. Add the directive
tls-auth-clients: optional(or"no") to keep accepting it. - Upgrade the operator first (its chart and CRDs), then change
spec.image: the operator rolls the pods replicas first, then fails the
master over, as for any engine upgrade.
kividb-operator v0.4.0
Added
KividbDbOpsCRD withop: restart/method: InPlacerolling
restart (replicas first, then master). GUI can create/list these.spec.bootstrapFromSnapshotonKividbCluster: seed a new
cluster's pod-0 PVC from a SucceededKividbSnapshotbefore the
StatefulSet starts (supported restore path; see
docs/BACKUP_RESTORE.md).- Agent
POST /execfor authenticated RESP commands (used by the GUI
explorer). - Agent
restore-from-s3subcommand used by the bootstrap Job. - Management GUI: Basic auth (
GUI_AUTH_*/ chartgui.auth.existingSecret),
live status gauges, pod logs, RESP explorer, DbOps restart actions,
24h local metrics scraper (PVC-backed) with sparkline charts. - e2e suite
08-resp-acl-select.sh: RESP3HELLO 3, ACL deny smoke,
SELECTon replica, failover single-master re-point. Override
KIVIDB_VERSION(e.g.v1.0.4-rc2) for pre-GA engine testing.
Changed
- Default engine pin
quay.io/kividbio/kividb:v1.0.5(and docs/
samples). Default agent image…-agent:0.4.0. v1.0.5 fixes all eleven
engine issues recorded indocs/KIVIDB_ENGINE_ISSUES.md, among them
unauthenticated access as thedefaultuser and the data-loss paths
around failed resyncs and skipped shutdown snapshots. - The agent sets
masterauthto thedefaultuser's password before
everyREPLICAOF. kividb v1.0.5 refuses an unauthenticated replication
handshake, so without it replicas of a password-protected cluster never
sync. It is set at runtime, so images of older engines keep working. - GUI ClusterRole gains
createonkividbdbopsandgetonpods/log
(still never Secrets to the browser). - Rolling updates are now done by the operator. The StatefulSet uses
theOnDeleteupdate strategy and the operator replaces pods itself:
replicas first, the master last, one at a time, and only while every
pod is Ready and every replica has finished syncing. A pod that is
unready on an outdated template is replaced without waiting. - The master is never restarted in place. A rolling update or
KividbDbOpsrestart first hands the master role to an in-sync replica
(eventSwitchover) and then replaces the old master. Restarting it in
place made every replica resync from whatever the master had last saved,
which on EKS lost 55 seconds of acknowledged writes during a config
change. Measured after the change: about one second of failed writes
and none lost. - Failover will not promote an empty replica when the master was last
seen holding data: kividb drops a replica's dataset when a full resync
fails part-way, which can empty every replica at once if the master is
crash-looping. The failover waits for the master instead (event
FailoverBlocked); set thekividb.io/allow-empty-failover: "true"
annotation on theKividbClusterto fail over regardless.
status.pods[].keysreports each pod's key count. status.pods[].syncedreports whether a replica has completed its full
sync from the current master.KividbDbOpsrestarts wait for it, and
failover prefers replicas that were in sync.
Fixed
- Failover could hand the cluster back to the old master and lose
writes. The pod being failed away from kept itsrole=masterlabel;
when it came back it rejoined the master Service, was picked as the
master again, and the pod that had been promoted was made its replica.
The old master is now relabeled before its replacement is promoted and
rejoins as a replica.status.phasealso no longer sticks at
FailingOverwithlastFailoverTimeadvancing every reconcile. - A master that had been turned into a replica was never repaired
(cluster reportedRunning, every write failed withREADONLY). The
operator now follows the pod it is replicating from if that is a healthy
master of the same cluster, and otherwise promotes it again. bootstrapFromSnapshotproduced an empty cluster. The restored
files were readable only by the restore Job's own user, so kividb could
not load them and started empty while the bootstrap reported success.KividbDbOpsrestart deleted every pod at once. A pod counted as
restarted while the deleted one was still terminating. Each pod must now
be replaced, Ready and back in the cluster before the next is deleted.- Backups failed with
archive/tar: write too longwhenever kividb
appended toappendonly.aofduring the upload. - ACL changes were never applied to running pods. The operator now
has each pod runACL LOADonce the updated file has reached it. KividbConfigchanges were never applied to running pods. They now
roll the pods, as does a change of thedefaultuser's password.- GUI: write endpoints (RESP explorer, restart, scale, promote, snapshot,
delete) were open when no auth Secret was configured, and accepted
cross-site requests. Promote sentREPLICAOFto backup Job pods, failed
halfway and could leave the cluster without a master. - The operator crash-looped if the
KividbDbOpsCRD was not installed,
which is the case after ahelm upgradefrom 0.3.0. - An unparseable
spec.storage.sizemade the reconciler panic in a loop;
it is now rejected by the CRD schema and reported on the cluster's
Readycondition. - The snapshot-restore pod was listed in
status.podsas a cluster
member, and pod-0's volume was the only one deleted with the cluster. - A rolling update required the pod being replaced to be in sync as well
as every other pod. Upgrading the engine from v1.0.4 to v1.0.5 stalled on
the last old pod: once the master role was on a v1.0.5 pod, the v1.0.4
pod could not authenticate to it and never became "synced". - A crash-looping pod blocked every rollout, including the template
change meant to fix it, and rollouts moved on to the master while the
replaced replica was still resyncing. See "Changed" below. - The GUI pod stayed
Pendingon clusters without a default StorageClass
(EKS): its metrics PVC was on by default and could not be provisioned.
gui.metrics.persistence.enablednow defaults tofalse. - Upgrading the operator while pods still ran kividb v1.0.3 never started
the rollout, because that engine does not report a replica's master. - The agent reported every replica's master port as 0, which made the
operator re-sendREPLICAOFto every replica on every reconcile.
Upgrade notes
- Apply the CRDs before upgrading the chart
(kubectl apply -f charts/kividb-operator/crds/):KividbDbOpsis new
andKividbClustergained validation. Helm does not do this for you. - Every existing cluster's pods are rolled once after the operator
is upgraded, one at a time: the pod template gains the agent's ACL
mount and thekividb.io/config-hash/kividb.io/auth-generation
annotations. - The GUI refuses all writes until
gui.auth.existingSecretis set. - Clusters without
spec.imagemove to kividb v1.0.5 in that same
rolling restart (from v1.0.3 if they were created by 0.3.0, or v1.0.4).
Pinspec.imagefirst to upgrade the engine separately. v1.0.5 changes
behaviour that worked before:- The
defaultuser's password is now enforced. Clients that never
sentAUTH(which worked, wrongly, on older engines) now get
NOAUTH. Check client configuration before upgrading. - Unknown or malformed directives in a
KividbConfigare fatal: the
pod will not start. Older engines ignored them silently. maxmemoryis enforced, withnoevictionby default: writes over
the limit are refused.
- The
- Clusters pinned to an engine older than v1.0.5 keep the old behaviour,
including unauthenticated access as thedefaultuser; restrict
network access to their Services.
Notes
- Known engine issue in v1.0.5: a replica whose CPU limit is changed in
place (for example by the Vertical Pod Autoscaler) panics on its next
full resync and stops replicating until restarted. The operator itself
never resizes pods in place. Seedocs/KIVIDB_ENGINE_ISSUES.md, issue 12. - Out of scope:
ReducedImpactrestart, in-place restore of a live
cluster.
kividb-operator v0.3.0
Added
- Multi-arch release images (
linux/amd64+linux/arm64) for the
manager, agent, and GUI. 0.2.0 published an OCI index with only amd64
(plus an attestation stub), which causedErrImagePull/
no matching manifest for linux/arm64/v8on Apple Silicon minikube and
other arm64 nodes. Release and CI Docker builds now use buildx with
QEMU andplatforms: linux/amd64,linux/arm64(provenance: falseso
the index is not polluted with unknown/unknown attestation-only
entries that confused some pullers). - Unit tests for core pure functions:
electReplica,computePhase,
renderKividbConf,renderACLFile, naming helpers, and
desiredBackupCronJob(go test ./... -race). - Minikube e2e harness under
hack/e2e/covering operator deploy,
kube-prometheus-stack, MinIO-backed snapshots, variant/TLS compat
against kividbv1.0.3, failover under load, snapshot chaos (pod kill
mid-backup), and agent memory metrics under load (make e2e).
Changed
- Samples, chart examples, and docs pin / recommend kividb v1.0.3
(andv1.0.3-tls/-lua/-fullvariants) as the validated engine
line for this operator release. Whenspec.imageis unset, the
operator now defaults toquay.io/kividbio/kividb:v1.0.3(no longer
the floating:latesttag). Defaultspec.agentImageis
quay.io/kividbio/kividb-operator-agent:0.3.0. - CI Go version bumped to 1.25 to match
go.mod/ Dockerfiles. - Dockerfiles honor
TARGETOS/TARGETARCHfrom buildx so multi-arch
compiles target the correct GOARCH.
Compatibility (kividb 1.0.3)
- TLS / Lua variants work on v1.0.3 (
v1.0.3-tls,v1.0.3-lua,
v1.0.3-full): e2e confirmed TLS portLISTENandEVALsuccess.
Prefer these tags over v1.0.2 for TLS. - Remaining upstream caveats (configfile
tls-*drop, ACL negative
command rules, unauthenticatedPING, replication RDB bulk-header)
still tracked in ROADMAP — re-check withhack/e2e/. - Snapshot robustness under source-pod or Job-pod kill is exercised by
hack/e2e/06-snapshot-chaos.sh(no resume of an in-flight
BGSAVE/upload; next manual/scheduled run should succeed after
recovery).
kividb-operator v0.2.0
Changed (breaking)
- Split the single
KividbClusterCRD into five (kividb.io/v1alpha1):
KividbCluster(topology, storage, scheduling, Services, monitoring,
failover),KividbConfig(reusablekividb.confdirectives + TLS
settings),KividbAclConfig(reusable ACL users /requirepass),
KividbSnapshotConfig(reusable backup schedule/destination/retention),
andKividbSnapshot(operator-created record of one backup run). A
StackGres-style split: configuration/ACLs/backup destinations are now
independent, reusable objects referenced by name
(spec.configRef/spec.aclConfigRef/spec.snapshotConfigRef) instead
of embedded fields.spec.auth,spec.backup, andspec.kividbConfig
no longer exist onKividbCluster. spec.imageis now the sole field determining which kividb image runs.
Removedspec.versionand the operator's automatic
quay.io/kividbio/kividb:v<version>[-<variant>]tag construction — the
operator never derives or modifies an image reference itself anymore.
Leavespec.imageunset for a floating, unpinned default tag; set it
explicitly to pin a version.spec.variant(standard/tls/lua/full) is now purely
informational: it tells the operator whether to wire up variant-specific
pod configuration (TLS cert mounting/CLI flags), but no longer
influences image resolution. A likelyspec.image/spec.variant
mismatch, or TLS enabled in aKividbConfigwithout a matching variant,
now surfaces as a guidance Event (VariantGuidance/
TLSVariantMismatch) viakubectl describe kividbclusterrather than
being silently wrong.KividbCluster.status.backupremoved — backup history now lives on
KividbSnapshotobjects (kubectl get kdbs -l kividb.io/cluster=<name>),
which carry per-run detail (source pod/role, object key, size, duration)
the old single status block couldn't.
Added
KividbSnapshotConfig.spec.source: master|replicato choose which
role's pod a scheduled backup is taken from (masteris the default
and recommended choice).values.schema.jsonfor the Helm chart.- Guidance Events (see above) surfaced on
KividbClusterreconcile. - Extensive rewrite of all docs for the new CRD architecture, plus
docs/ROADMAP.md(0.2.0/1.0.0 planning and known upstream kividb
issues found via live testing) anddocs/VERSIONING.md(how an
external site fetches these docs at a specific released version via
git refs).
Fixed
- TLS settings (
tls-port/tls-cert-file/tls-key-file/
tls-ca-cert-file) are now also passed to thekividbcontainer as CLI
flags, not only written into the renderedkividb.conf— live testing
found kividb's--configfileparser currently ignores these directives
when file-sourced, even though the identical CLI flags are documented
and accepted.
Security
- Bumped the builder image for all three published images
(kividb-operator,kividb-operator-agent,kividb-operator-gui) from
golang:1.23-bookwormtogolang:1.26-bookworm, and bumped
golang.org/x/net(0.28.0→0.56.0),golang.org/x/oauth2
(0.21.0→0.27.0),golang.org/x/sys(0.24.0→0.46.0), and
golang.org/x/text(0.17.0→0.39.0) — addressing a batch of stdlib and
dependency CVEs (including a criticalcrypto/tlssession-resumption
issue, CVE-2025-68121) reported against the0.1.0images.