FXVPS
Get Started
vps expert-advisors redundancy risk-management mt4 mt5

A Backup VPS for Your EAs: How to Run a Standby Without Doubling Every Trade

person
FXVPS
Share
A Backup VPS for Your EAs: How to Run a Standby Without Doubling Every Trade

Once an account is large enough, or a funded account is valuable enough, many traders ask the same question: what if my VPS goes down? A good datacenter makes that rare. Redundant power and networking are exactly why a VPS beats a home PC. But rare is not never, and a single machine is still a single point of failure.

The obvious answer is a second VPS. The obvious way to set it up is to run the same EA on both. That is also the most expensive mistake in this whole topic. MetaTrader will happily let two terminals log in to one account at the same time, and two copies of an EA that each see a signal will each place the trade. Your risk doubles and nothing tells you.

Redundancy for EAs is a real technique, but it is not the same as redundancy for a website. Here is how to do it properly.

📊 Key Stat: MT4 and MT5 generally allow the same trading account to be logged in from multiple terminals at the same time. The broker’s server treats orders from each terminal as independent requests. Nothing stops two EAs on two machines from opening the same position twice.

Why “Just Run It Twice” Fails

Consider a simple trend EA with one rule: open a buy when the fast moving average crosses above the slow one, if no buy is open.

Run it on two VPSs against the same account, and both terminals see the cross on the same tick, or near enough. Both check “is a buy open?” Neither has seen the other’s order yet, so both answer no. Both send. You now hold twice the intended position. Whether the EA’s later logic manages two positions sensibly depends entirely on how it was written.

It gets worse with grid, martingale and basket EAs, which track their own state. Two copies each build their own view of the grid, and the account ends up with a structure neither copy intended. See our grid and martingale guide for why these systems are sensitive to exactly this.

⚠️ Warning: Magic numbers do not protect you here. Both copies of the EA use the same magic number, so each can treat the other’s trades as its own. That can make things worse: one copy may close or modify positions the other just opened.

The Three Standby Models

Cold Standby: Ready but Not Logged In

The second VPS has MetaTrader installed, your EAs and set files copied, and charts laid out, but the terminal is not running (or not logged in to the account).

  • Failover: You notice the primary is down, RDP into the standby, launch the terminal, log in and enable AutoTrading.
  • Risk of double trading: None while the standby is cold.
  • Takeover time: Minutes, depending on how fast you notice.

This is the right choice for most traders. It removes the worst part of a VPS failure, which is scrambling to rebuild a setup from scratch, and adds no new risk.

Warm Standby: Logged In, EA Disabled

The standby terminal runs and is logged in with the EA attached to its charts, but AutoTrading is off, or the EA’s own “allow trading” input is false.

  • Failover: Flip AutoTrading on. That takes one click.
  • Risk of double trading: Low, provided AutoTrading truly stays off. The failure mode is a terminal restart or a settings change that re-enables it without you noticing.
  • Takeover time: Seconds after you decide to act.
  • Bonus: The standby keeps its history and chart data current, so indicators are warmed up when it takes over.

💡 Tip: On a warm standby, add a visual signal you cannot miss, such as a chart background colour or a large text label reading “STANDBY – DO NOT ENABLE”. The dangerous moment is a tired trader connecting to the wrong machine and “fixing” what looks like a disabled EA.

Hot Standby: Automatic Failover

Both machines run the EA, and some coordination mechanism lets only one trade at a time. The standby watches for the primary’s heartbeat and takes over when it stops.

This is how institutional systems do it, but it is hard to build correctly for MetaTrader EAs:

  • The coordination has to live outside both VPSs. A heartbeat file or terminal global variable on the primary vanishes with the primary.
  • “The primary looks down” and “the primary is down” are different things. If the network between the machines blips, both may decide they are primary. That is split-brain, which leads straight back to double trading.
  • The EA itself must support it, and most commercial EAs don’t.

Unless you or your developer are building this deliberately, with the heartbeat service and the split-brain handling included, don’t improvise a hot standby. A cold or warm standby with good alerting covers most of the benefit at a fraction of the risk.

Make Your EA Failover-Friendly

Whichever model you choose, the takeover goes smoothly only if the EA on the standby can pick up where the primary left off.

  • State from the account, not from files. An EA that rebuilds its view on startup from open positions and orders (filtered by magic number) takes over cleanly. An EA that keeps its state in a local file (MQL4/Files or MQL5/Files) or in terminal global variables starts the standby with no memory of what the primary was doing.
  • Idempotent entries. Before opening a trade, a well-behaved EA checks whether the position it intends to open already exists. This one habit also protects against duplicate orders after timeout errors.
  • Same inputs, same version. Use the same set files and the same EA build on both machines. Keep them in sync every time you change a parameter, or the standby takes over with last month’s settings.

✅ Best Practice: Test the failover before you need it. On a demo account, run the primary, stop it mid-trade, bring the standby up and watch what the EA does with the existing positions. If it opens new ones or ignores the old ones, you have found the problem on demo rather than live.

Where to Put the Standby

Same region as the primary keeps latency identical after failover, so your EA’s execution does not change during an incident. Choose this if latency matters to your strategy.

A different region protects against a regional event, at the cost of a latency penalty while the standby is active. For example, a London primary with a New York standby for a broker whose servers are in London means slower execution during the outage, but you still manage your positions. For most swing and position strategies that trade-off is fine. For scalpers it may not be. Our guide to moving VPS locations explains how much the latency changes.

Your home PC can act as a cold standby for smaller accounts. It is free, but it only helps if it is actually powered on and up to date when you need it.

⚠️ Warning: Prop firm traders should read their firm’s rules on IP addresses before setting up a standby. Some firms watch for logins from multiple IPs or regions. A failover that looks like account sharing can cause more trouble than the outage. See multiple prop firm accounts and IP rules.

Know When to Fail Over

A standby only helps if you know the primary is down. Set up:

  • MetaTrader push notifications from the EA, including a periodic “alive” message if the EA supports it. Silence becomes the alarm. See push notifications and monitoring.
  • An external uptime check that pings the primary VPS and alerts your phone.
  • A written runbook: which machine, which login, which button. At 3am you should not be working it out.

Before failing over, confirm the primary is really down, not just that your RDP session failed. A primary that is still trading plus a standby you just enabled is the double-trade scenario from the start of this article. Our Remote Desktop errors guide covers how to tell the two apart.

🚀 Try FXVPS for $2.99. Set up a standby on a 7-day Core VPS trial and rehearse a failover on demo before you rely on it.

What It Costs

A cold standby does not need to match the primary’s size. Its job is to run your terminals for hours or days, not to replace your whole operation permanently. For many traders a Core plan ($29/mo, 2GB RAM, 1 vCPU) is a sufficient standby, even behind a Pro ($39/mo) or Scaling ($79/mo) primary, as long as it can run the terminals that matter. Size it for the accounts you would actually fail over, not for everything. See pricing for current plans.

Is it worth it? Compare the monthly cost to one bad outcome: an unmanaged position through a news event, or a funded account breached because a stop never trailed. For a small personal account, a good primary VPS plus a documented rebuild process is probably enough. For a funded account or managed capital, a standby is cheap insurance.

Frequently Asked Questions

Can I run the same EA on two VPSs for safety?

Not both actively on the same account. Two active copies will usually double your trades. Use a cold or warm standby, where only one copy can trade at a time.

Will my broker block a second login to my account?

Usually not. MT4 and MT5 generally allow simultaneous logins, which is exactly why double trading is a risk. Some brokers and prop firms have their own policies, so check them.

Does the standby need the same plan as my primary VPS?

No. It needs enough resources to run the terminals you would fail over. A smaller plan is often enough for a standby.

What about my open trades if the primary VPS dies?

They stay on the broker’s server, with their stop losses and take profits intact, because those are stored server-side. What stops is anything the EA does actively: trailing, partial closes, virtual (hidden) stops and new entries. Our article on what stops working when your terminal goes offline covers this in detail.