ezgh caches get cch_4k2m9x0a7qpeThe console shows the same on the cache’s Connectivity tab: the endpoint, its ports, and the
default user’s Secret.
- Endpoint: the cache’s DNS name. It stays the same through failovers, maintenance, upgrades and blue/green switchovers.
- Port: 6379 for Valkey. 11211 (TLS) and 11212 (no TLS) for Memcached.
- User and password: in each user’s Secret. On Valkey, its
urifield is a ready-maderediss://URL.
Connect from an address in the cache’s allowed sources, inside its VPC.
TLS
Valkey takes only TLS connections. The cache’s certificate is for its endpoint and chains to the EZGH CA bundle. Verify the server’s certificate and its name against the bundle:
curl -fsSO https://get.ezghcloud.com/caches/ca-bundle.pemvalkey-cli --tls --cacert ca-bundle.pem -h cache-hylq6kelvuwvn2ks4yqhp4jaqu.us-west-1.ezghcompute.com \
-p 6379 --user default --askpassredis-cli takes the same options.
import redis
r = redis.Redis(
host="cache-hylq6kelvuwvn2ks4yqhp4jaqu.us-west-1.ezghcompute.com",
port=6379,
username="default",
password=password,
ssl=True,
ssl_ca_certs="ca-bundle.pem",
socket_connect_timeout=2,
socket_timeout=2,
)import { readFileSync } from "node:fs";
import Redis from "ioredis";
const redis = new Redis({
host: "cache-hylq6kelvuwvn2ks4yqhp4jaqu.us-west-1.ezghcompute.com",
port: 6379,
username: "default",
password,
tls: { ca: readFileSync("ca-bundle.pem") },
connectTimeout: 2000,
commandTimeout: 2000,
});pool := x509.NewCertPool()
pool.AppendCertsFromPEM(bundle)
rdb := redis.NewClient(&redis.Options{
Addr: "cache-hylq6kelvuwvn2ks4yqhp4jaqu.us-west-1.ezghcompute.com:6379",
Username: "default",
Password: password,
TLSConfig: &tls.Config{RootCAs: pool, ServerName: "cache-hylq6kelvuwvn2ks4yqhp4jaqu.us-west-1.ezghcompute.com"},
DialTimeout: 2 * time.Second,
ReadTimeout: 2 * time.Second,
WriteTimeout: 2 * time.Second,
})Memcached clients connect with a username and password over the binary protocol (SASL): with TLS
on port 11211, or without TLS on port 11212 while the cache’s plaintextPort is on. Turn it off when
every client uses TLS:
ezgh caches update cch_4k2m9x0a7qpe --plaintext-port=falseTimeouts
After a failover or maintenance, the endpoint points somewhere new. An open connection doesn’t follow it: it ends only when it fails or times out, however short the DNS TTL is. Set a connect timeout and a read timeout of a few seconds in every client, and let the client reconnect to the endpoint when a connection fails. Pooled connections should be tested or replaced when they fail.
DNS
The endpoint is a DNS name for a private address in your VPC. Its record has a 5-second TTL. Set clients to cache it for no longer:
- Java: a JVM caches names for 30 seconds by default. Set
networkaddress.cache.ttl=5, in the JDK’sconf/security/java.securityor withjava.security.Security.setProperty("networkaddress.cache.ttl", "5")before the first connection. - Your own caches: a resolver cache on the instance, or a connection string with an address in place of the endpoint, has the same effect. Use the endpoint, and honor its TTL.
Instances resolve the endpoint through public DNS, so an app’s instance needs a path to the internet to look it up, for example through a Network Gateway or a public IPv4 address. The connection itself stays inside the VPC.
Rate limiting
A fixed-window limiter counts requests per key with INCR and starts the window with EXPIRE:
INCR ratelimit:api:203.0.113.7
EXPIRE ratelimit:api:203.0.113.7 60 NXAllow the request while the count is at or below the limit. To do both in one round trip and atomically, run them as a script:
local n = redis.call("INCR", KEYS[1])
if n == 1 then redis.call("EXPIRE", KEYS[1], ARGV[1]) end
return nlimit = r.register_script(SCRIPT)
allowed = limit(keys=[f"ratelimit:api:{client_ip}"], args=[60]) <= 100Users with the readWrite preset can run scripts (EVAL, EVALSHA, FUNCTION) and use pub/sub.
Decide what your app does when the cache can’t be reached, such as allowing the request, and keep the
client’s timeouts short so a slow answer doesn’t hold requests up.