Skip to main content

Install Spider with one command

New script

The installer is new. It has been tested against a stubbed Floocus Portal and cluster, not yet widely in the field. For a production setup, prefer the manual path (browser wizard).

One command installs a complete Spider on a Linux host (or on a Kubernetes cluster you can reach), with a free BASIC licence. You can run it yourself, or give it to an AI agent (see the spider-install skill of the Spider Claude plugin).

curl -fsSL https://spider-analyzer.io/install.sh | bash -s -- --email you@example.com --accept-terms --install-k3s

Pipe into bash, never into sh: the script needs bash.

Add --dry-run first to see what would be done without changing anything.

What is asked of the human​

  • Your email address. It becomes your Floocus account and the Spider admin login.
  • Agreement to the Floocus terms, given with --accept-terms. Nobody should pass this flag on your behalf.
  • One code. Floocus emails a 6-digit one-time code (OTP) to that address, valid 10 minutes. It is sent only during the real run, so you cannot have it before the first run. In a terminal, the script asks for it and you type it. Without a terminal (an agent, a pipeline), the flow is:
    1. Run the command without the code. The script registers the account, Floocus sends the code, and the script stops with exit 30 and the message no code: a 6-digit code was emailed to <email>.

    2. Read the code in your mailbox.

    3. Run the same command again with the code in the environment of bash, not of curl. The variable goes after the pipe:

      curl -fsSL https://spider-analyzer.io/install.sh | SPIDER_INSTALL_OTP=<code> bash -s -- --email you@example.com --accept-terms --install-k3s

      Writing the variable before curl sets it on curl only: the script sees no code and stops again with exit 30. The run resumes once it has the code. If the code expired, the script requests a new one.

Prefer the prompt or the environment variable for the code

--otp <code> works, but a command line is visible to the other users of the machine in the process list while the script runs. Use SPIDER_INSTALL_OTP or the prompt. If the email already has a Floocus account, the script needs its password instead: type it when asked, or set SPIDER_INSTALL_FLOOCUS_PASSWORD (environment only, never a flag), again on bash: curl -fsSL https://spider-analyzer.io/install.sh | SPIDER_INSTALL_FLOOCUS_PASSWORD=<password> bash -s -- --email you@example.com --accept-terms.

A new account's organisation is named after the domain of your email address (gmail.com for a Gmail address), and that name is stored by Floocus. To choose it, set SPIDER_INSTALL_ORG_NAME the same way, on bash: curl -fsSL https://spider-analyzer.io/install.sh | SPIDER_INSTALL_ORG_NAME="Acme" bash -s -- --email you@example.com --accept-terms.

Flags​

FlagMeaning
--email <address>Floocus account email, also the Spider admin login
--otp <code>The 6-digit code emailed by Floocus (or SPIDER_INSTALL_OTP; else typed on the terminal)
--accept-termsAccept the Floocus terms, needed to register
--host <name>Spider host name (default spider.<node-ip>.nip.io)
--install-k3sInstall k3s when no cluster is reachable
--ingress-class <name>Ingress class (default: detected traefik/nginx; traefik on k3s)
--storage-class <name>StorageClass for Elasticsearch (default: the cluster default)
--namespace <name>Spider namespace (default spider-system)
--rotate-licenceRe-issue the credentials of the organisation's existing licence when they are not on this host. Every other install of that licence stops working. Asks first, or needs --yes
--yesNon-interactive: never prompts. Confirms --rotate-licence. Never installs k3s on its own (pass --install-k3s)
--verify-fullAlso deploy spider-hello and check each protocol it speaks (HTTP, TLS, h2, gRPC, SSE, WebSocket, DNS). Failures are reported, the install still succeeds
--jsonJSON summary on stdout, messages on stderr
--dry-runShow what would be done, change nothing
-h, --helpUsage

What gets created​

  • A Floocus account and organisation, and a BASIC licence for it, through the Floocus Portal.
  • k3s, only with --install-k3s and only when no cluster is reachable.
  • The ECK operator (Elasticsearch), then the Spider Helm chart in the namespace spider-system.
  • A default whisperer, seeded by the chart.
  • A spider CLI profile named spider-install, and the spider binary itself when none is on the PATH (~/.local/bin, or /usr/local/bin as root).
  • With --verify-full: the spider-hello demo workload.

The script ends by checking that Spider captures its own HTTP traffic. An active Spider alert fails the install (exit 70); an unreachable /health is only a warning.

Which address Spider gets​

Unless you pass --host, the host name is spider.<ip>.nip.io, where <ip> is the first public IPv4 address of the cluster node (or of the host), else a private one. If only a private address is found, your browser must be able to reach it; otherwise pass --host. The URL is plain HTTP: it is a lab install, with no certificate. For HTTPS, use the manual path.

Where the secrets live​

Everything is in ~/.spider-install/ (mode 600 files, 700 directories): the state of the run, secrets/setup.yaml, keys and the admin password (secrets/cluster/admin.password). The script never prints a secret, only the paths. Keep that directory: it is the only copy that lets a later run keep your licence.

Re-running​

The script is idempotent. A re-run resumes from its state file and only changes what differs. It never rotates the licence credentials by itself: if the organisation already has a licence and the credentials are not on this host (state lost, other machine), it stops with exit 30 and names --rotate-licence. Exit 30 also covers a missing code (see above), a missing password for an existing account, and the sign-up limits below. Use that flag only when you accept that every other install of the same licence stops working. The terms agreement is not remembered: pass --accept-terms again on a re-run that needs to register.

Exit codes​

CodeMeaning
0Done
2Usage: invalid option, or not run with bash
5Bootstrap refused: download or checksum failed, nothing was run. Retry in a few minutes
10Preflight failed (tools, OS, privileges, host)
20Cluster: none reachable, or k3s install failed
30Account: Floocus account, organisation or licence
40Setup: cluster secrets or setup.yaml
50ECK: Elasticsearch operator CRDs
60Install: helm install of the Spider chart
70Verify: login, CLI, capture or an active Spider alert

With --json, stdout carries one document: status, phase, exitCode, message, dryRun, host, url, namespace, email, licenceId, setupFile, adminSecretsFile, flipsOff (the cluster flips left off) and, with --verify-full, verifyFull. It holds paths and ids, never a secret.

Integrity of the download​

install.sh downloads its files from https://spider-analyzer.io/install/ and checks them against SHA256SUMS, served from the same site. That detects a corrupted or truncated download. It does not protect against a compromised site: whoever can alter the script can alter the checksums. To audit first, download install.sh, read it, then run it with bash install.sh.

Limits​

  • Account creation through this path is capped, to bound abuse: 5 confirmed accounts per registered email domain (lifetime; webmail domains such as gmail.com are exempt from this one), 5 registrations per IP address per day, 100 registrations per day for everyone, and one organisation per email address. These are the values in the Floocus code today and may change. Beyond a cap, the script stops with exit 30 and asks you to contact support@floocus.com (quote the email you used). A Floocus service outage answers 503 instead: retry later.
  • A BASIC licence limits some features: see the licence tiers.
  • The installer is new; report problems to support@floocus.com.

Undo​

Uninstalling Spider destroys your captured data

helm uninstall removes the Elasticsearch resource, and by default ECK then deletes its volumes: the captured traffic is gone. The chart does not set volumeClaimDeletePolicy (the only volumes it creates are Elasticsearch's), and this is ECK's default behaviour; we have not tested it on a cluster. There is no safe first step: if you care about the data, export or back it up before running any command below.

helm uninstall spider -n spider-system # destroys the Elasticsearch data
kubectl delete namespace spider-system # removes what is left of the namespace
kubectl delete namespace spider-hello # only if you used --verify-full

What remains, and how to remove it:

  • ECK CRDs. They are cluster-scoped and stay. Remove them only if the installer applied them (eckCrds is not empty in ~/.spider-install/state.json) and nothing else on the cluster uses ECK: deleting a CRD deletes every Elasticsearch, Kibana and other ECK object of the cluster. Then: kubectl get crd -o name | grep k8s.elastic.co | xargs kubectl delete.
  • k3s. Only if this script installed it and you want the cluster gone: /usr/local/bin/k3s-uninstall.sh. It wipes the whole cluster and every workload on the host, not only Spider.
  • On the host. The spider-install profile in ~/.spider/profiles.json, the spider binary the script installed (~/.local/bin/spider, or /usr/local/bin/spider as root), the Helm repository spider-floocus (helm repo remove spider-floocus), and ~/.spider-install/. Delete that directory only when you no longer need the licence credentials.
  • At Floocus. The account, organisation and licence remain: these commands do not touch them. To close or delete them, write to support@floocus.com.

Next​

Sign in at the printed URL with your email and the password in ~/.spider-install/secrets/cluster/admin.password, then start a capture.