Help
Valkey and Redis errors and how to fix them
Tested with:Databasezy API v1 · zb CLI 0.1
Valkey is available on every plan, including Free. Redis (compat), a Redis 7.2 server, is available on paid plans only.
Both speak RESP, accept TLS connections only, and use the default user with a generated password. Gateway errors
arrive as normal -ERR replies, so your client shows them like server errors.
Connection reset, timeout or “Connection closed” right away
Section titled “Connection reset, timeout or “Connection closed” right away”Cause: in these cases the gateway closes the connection without sending a RESP reply:
- The client did not use TLS. For example, a
redis://URL orvalkey-cliwithout--tls. The port is TLS-only, and a plaintext client gets no reply. - The client did not send the host name (SNI). For example, you connected by IP address. The gateway picks your instance from the TLS server name.
- Your address is not on the allow-list, or it is temporarily blocked after repeated failed logins. By default, 10 failures within a minute block the address for 10 minutes.
Fix: use rediss:// (two s) with the host from the Connect tab, keep hostname verification on, and add your
egress CIDR under Network → IP allow-list. See Connect to Valkey and
IP allow-lists.
WRONGPASS invalid username-password pair or user is disabled
Section titled “WRONGPASS invalid username-password pair or user is disabled”Cause: the password is wrong, or it was rotated. After a rotation the server restarts, so every client must reconnect.
- Valkey: the old password keeps working until 10 minutes after the rotation started.
- Redis (compat): the old password stops working when the server restarts.
Fix: reveal the current password (zb instances credentials reveal <instance_id>). If your client asks for a
username, use default. See Rotating credentials.
NOPERM or ... command not allowed
Section titled “NOPERM or ... command not allowed”Cause: Databasezy manages the server, so default cannot run commands that change configuration, persistence,
replication, modules or users. Examples are CONFIG SET, SAVE, BGSAVE, REPLICAOF, MIGRATE, MONITOR,
SHUTDOWN, CLIENT PAUSE and ACL SETUSER. DEBUG and MODULE LOAD are refused with ... command not allowed.
The same rules apply inside Lua scripts.
Fix: remove the command from your startup code. Common cases:
- Spring Session runs
CONFIG SET notify-keyspace-eventsat startup. Setspring.session.redis.configure-action=none, then contact support to setnotify-keyspace-eventsas a managed setting. - You need a different
maxmemory-policy,notify-keyspace-events,appendfsyncortimeout. You cannot change these from the portal or API yet; contact support. A change restarts the instance once.
The full list is in Valkey security.
OOM command not allowed when used memory > 'maxmemory'
Section titled “OOM command not allowed when used memory > 'maxmemory'”Cause: the dataset reached maxmemory, which is 75% of the size’s memory. The policy is noeviction, so writes
fail while reads and deletes still work. Keys are never evicted.
Fix: delete keys or set TTLs on cache keys, or resize. Watch used_memory in INFO memory.
MISCONF errors and refused writes
Section titled “MISCONF errors and refused writes”Cause: the disk is full, so a snapshot or append-only file write failed.
Fix: grow the instance’s storage. See Resize.
ERR max number of clients reached
Section titled “ERR max number of clients reached”Cause: the instance already has as many connections as its size allows. Each pub/sub subscriber and each blocking
command (BLPOP, XREAD BLOCK) holds a connection of its own. ERR connection rate limit exceeded; slow down means
clients opened more than 50 new connections in one second.
Fix: use one multiplexed client per process instead of a pool, and count your subscribers. See Too many connections.
ERR paused: ..., ERR the instance is waking up ... or ERR suspended: ...
Section titled “ERR paused: ..., ERR the instance is waking up ... or ERR suspended: ...”Cause:
pausedmeans someone paused the instance on purpose; connections never wake it. Runzb instances resume <instance_id>. If the text sayspaused by Databasezy support, only support can resume it.waking upmeans a paused instance did not wake within the gateway’s 30-second hold. Free instances pause after 15 minutes without connections. A client that reconnects on error handles this.suspended(orquota_exceeded) means billing or a quota stopped the instance.
Fix: see Paused and waking instances and Spending caps and quota errors.
ERR unknown command 'JSON.SET' (or FT.*, BF.*)
Section titled “ERR unknown command 'JSON.SET' (or FT.*, BF.*)”Cause: these commands come from modules.
- Valkey: turn on
valkey-json,valkey-bloomorvalkey-searchper instance, on Valkey 8.1 or later. Turning one on restarts the instance once. - Redis (compat): no modules are loaded. The extensions API answers
409.
Fix: see Valkey modules. Redis Stack modules are not offered.
Bad data format after RESTORE
Section titled “Bad data format after RESTORE”Cause: RESTORE payloads are fully checked before they load, so a malformed or incompatible DUMP value is
refused.
Fix: dump the key again from the source, or use zb migrate, which copies keys for you.
Data is missing after a restore
Section titled “Data is missing after a restore”Restores return to one RDB snapshot. There is no point-in-time recovery for these engines, so writes made after that backup are not in it. See Restore a backup.
Still stuck?
Section titled “Still stuck?”Search the help center or ask the assistant in the portal Help panel. If that does not solve it, open a support ticket from the portal (Help → Open a ticket) and include:
- the instance id, engine (Valkey or Redis) and region
- the exact error text
- your client library and its version
- the command that failed
- the time of the failure in UTC
Never paste passwords, connection strings with passwords, API keys or patient data (PHI) into a ticket. See How to get help.