Sessions & rotation

A "session" is a way to keep the same residential exit IP across multiple requests. helodata supports two modes; you pick per request via the username.

Rotating mode (default)

Omit the session and sesstime segments. Every new TCP connection picks a fresh IP from the matching pool.

helo_s1a2b3c4d5e-type-res-region-us

When to use:

• One-off page fetches that don't need cookies
• Anti-bot evasion via IP diversity
• Maximum pool diversity for large parallel crawls

Caveat: connection reuse (keep-alive) within a single client may pin you to one IP until the connection closes. If you want literal-per-request rotation, force fresh connections (Connection: close).

Sticky mode

Append both session-{id} and sesstime-{minutes}. Requests sharing the same combined username re-use the same exit IP for up to sesstime minutes.

helo_s1a2b3c4d5e-type-res-region-us-session-job1-sesstime-30

Constraint	Value
Max sesstime (residential)	30 minutes
Session ID character set	[a-z0-9], max 32 chars
IP guarantee	best-effort — sessions can be re-pinned if the upstream IP drops
Session expiry	first request starts the timer

When to use:

• Multi-step flows (login → search → checkout)
• Sites that bind anti-bot signals to IP across a few minutes
• Tests where you need response reproducibility

Concurrency under sticky

Two requests with the same session_id but different other parameters (different geo, different sub-user) are treated as separate sessions. The whole username string is the session key, not just the ID.

So this:

helo_s1a2b3c4d5e-type-res-region-us-session-abc-sesstime-30
helo_s1a2b3c4d5e-type-res-region-de-session-abc-sesstime-30

…opens two unrelated sessions, one in the US and one in DE.

Recovering from a dropped session

If your session's upstream IP drops mid-flight, the next request with the same username returns a fresh IP and the timer restarts. You'll see this in the response header:

X-Helodata-Session-Repinned: true

Watch for this header in long-running flows and consider re-establishing client state (cookies, fingerprint) when it appears.

Patterns

Pool of N concurrent sessions

import requests, itertools, random, string

BASE = "helo_s1a2b3c4d5e-type-res-region-us"

def session_user(n):
    sid = "".join(random.choices(string.ascii_lowercase + string.digits, k=8))
    return f"{BASE}-session-{sid}-sesstime-10"

users = [session_user(i) for i in range(20)]      # 20 sticky sessions

def get(url, user):
    proxy = f"http://{user}:[e-mail]:7777"
    return requests.get(url, proxies={"http": proxy, "https": proxy}, timeout=30)

Rotate users round-robin across your workers.

Burn-and-replace

Track failed requests per session ID. When error rate per session crosses a threshold (e.g. 3 failures in 60s), discard that session ID and mint a new one.

What sessions do not do

• They don't extend traffic credits — bandwidth is still billed per byte.
• They don't carry cookies, headers, or TLS fingerprints — that's your client's job.
• They don't guarantee a specific IP — only IP stability for the session window.

Related

• ASN targeting — combine with sessions to keep the same ISP throughout
• Locations — geo segments must match across requests to share a session
• 407 errors — common if session syntax is malformed