Asama

Asama plans the fix. You approve it.

For each incident, Asama writes a node-specific fix and the code that verifies it, following your maintenance policy. Nothing runs until you approve it. Rollback is ready before the first step, and every run improves the next plan.

Read-only by default · No training on your data · SOC 2 Type II

How the remediation loop works

Detection and diagnosis hand over the root cause and its evidence. From there it’s one loop: Asama plans, you approve, the workflow runs and verifies, and every run improves the next plan.

Your fleet is heterogeneous. The same server model runs different firmware and BIOS versions from one rack to the next, so a single automation never fits every host. That leaves your team maintaining runbooks and scripts for every variation, and rewriting them each time something changes.

A plan, not a suggestion

Asama starts from the diagnosis, the root cause and its evidence, and plans for this incident on this node, using its live configuration, firmware and versions, and what worked in past runs. Each plan is sequenced into phases with the exact changes and commands.

  • Writes the fix and the code that verifies it
  • Plans for each node’s firmware and BIOS
  • Success and failure paths set before it runs
Config Files lab-sf-db07

From 3 past db-tier journald fixes

Awaiting approval

Planned from

Root cause

Journal 1.9G, disk 91% full

Node state

systemd 252, no drop-ins

Fix

3. Change journald.conf:
− SystemMaxUse=500M
+ SystemMaxUse=200M

Verification · runs after each batch

systemd-analyze cat-config = 200M
journalctl --disk-usage ≤ 200M

Built against your policy

Maintenance windows, batch limits and environments are inputs to the plan, not something you apply afterwards.

  • Respects your windows and batch limits
  • Holds approved plans for the next window
  • Announces the change to workload owners
Config Files lab-sf-db05–db09

Held by the maintenance orchestrator

Scheduled

Maintenance policy

Window

Tue 14 Oct · 02:00–04:00 PDT

Weekly db-tier window

In window
Batch

2 hosts at a time

db-tier batch limit

Within limit

Workload owners

postgres · pg-orders-replica-2

lab-sf-db07 · Data Platform

Notified
Runs Tue 14 Oct, 02:00 PDT once approved

You approve, or you change it

Nothing runs until you approve it. Approve the plan as it stands, or edit the code and ask Asama to revise it and explain its choices.

  • Flags risks and policy conflicts
  • Saves your edits for the next plan
  • Run it now, or schedule it
Config Files lab-sf-db07

1 risk flagged · 1 edit

Awaiting approval

Risk flagged

Restarting journald briefly pauses log shipping.

Prepare

2. Wait 30 minutes before next remediation steps.
Edited · was 10 min Saved for future db-tier plans.
Nothing runs until approved

Executed, verified, escalated

A durable execution workflow applies the fix in batches and runs the verification code after each one. Rollback is computed before the first step.

  • Rolls back on a failed check
  • Alerts you and brings a revised plan
  • Each run improves the procedure library
Config Files lab-sf-db05–db09

5 hosts · batch size 2

Halted · re-planned

Hosts

lab-sf-db05

Verified

Journal at 186.0M.

02:11

lab-sf-db06

Rolled back

Check failed at 488.0M. Rolled back.

02:17
Asama re-planned · 02:18 PDT

A vendor drop-in overrides SystemMaxUse. The revised plan waits for your review.

Review revised planAbort run

FAQ

See Asama on your own fleet

Bring one recurring problem your current tooling hasn’t solved. We’ll show you how Asama detects it, finds the cause, and plans the fix.