⚠️ *Heads-up for RPC node operators on kaswallet v3.0.2 and older — protect your wallet keys*
*Who this is for:* anyone running an IGRA node with *kaswallet v3.0.2 or earlier* that serves *RPC and submits transactions* — i.e. your worker wallets are *funded*. Those are the wallets most worth protecting here.
*The issue (rare, but it can cause key loss):*
kaswallet ≤ v3.0.2 rewrites keys.json *in place* on every address/index update. In an uncommon timing window, if the process is killed *during* that write (e.g. a container/dependency restart landing mid-save), the file can be left *truncated to 0 bytes*. Without a backup or your mnemonic, that wallet — *and any funds in it* — become *unrecoverable*. It's rare, but funded RPC wallets write keys.json often, so they're the most exposed.
*① Check your kaswallet version* — it's the Docker image tag. From your igra-orchestra directory:
grep KASWALLET_VERSION .env # the version you've pinned
docker compose images | grep kaswallet # the image your workers are actually running
• *3.0.2 or lower* ➜ affected — follow the steps below.
• *3.0.3 or higher* ➜ you already have the fix (just make sure you've done the key-directory migration).
• *3.0* (a floating tag) ➜ could be either; treat as affected and pin KASWALLET_VERSION=3.0.3 explicitly to be sure.
*② Back up your keys now — do this regardless of version.*
An encrypted, off-box backup of your keys/ directory (or at least your mnemonics) fully protects you against this bug today.
tar czf keys-backup-$(date +%F).tgz keys/ # then move it off the box and encrypt/store it safely
*③ Upgrade to kaswallet v3.0.3 — the permanent fix.*
v3.0.3 writes keys.json *atomically* (temp file ➜ fsync ➜ rename), so an interrupted save can never truncate your key file again. The atomic write needs a per-worker *directory* mount, so the upgrade includes a one-time key migration. On each node:
⚠️ *Stop your frontend workers FIRST*, before you migrate or upgrade. Nothing must be
⚠️ *Heads-up for RPC node operators on kaswallet v3.0.2 and older — protect your wallet keys*
*Who this is for:* anyone running an IGRA node with *kaswallet v3.0.2 or earlier* that serves *RPC and submits transactions* — i.e. your worker wallets are *funded*. Those are the wallets most worth protecting here.
*The issue (rare, but it can cause key loss):*
kaswallet ≤ v3.0.2 rewrites keys.json *in place* on every address/index update. In an uncommon timing window, if the process is killed *during* that write (e.g. a container/dependency restart landing mid-save), the file can be left *truncated to 0 bytes*. Without a backup or your mnemonic, that wallet — *and any funds in it* — become *unrecoverable*. It's rare, but funded RPC wallets write keys.json often, so they're the most exposed.
*① Check your kaswallet version* — it's the Docker image tag. From your igra-orchestra directory:
grep KASWALLET_VERSION .env # the version you've pinned
docker compose images | grep kaswallet # the image your workers are actually running
• *3.0.2 or lower* ➜ affected — follow the steps below.
• *3.0.3 or higher* ➜ you already have the fix (just make sure you've done the key-directory migration).
• *3.0* (a floating tag) ➜ could be either; treat as affected and pin KASWALLET_VERSION=3.0.3 explicitly to be sure.
*② Back up your keys now — do this regardless of version.*
An encrypted, off-box backup of your keys/ directory (or at least your mnemonics) fully protects you against this bug today.
tar czf keys-backup-$(date +%F).tgz keys/ # then move it off the box and encrypt/store it safely
*③ Upgrade to kaswallet v3.0.3 — the permanent fix.*
v3.0.3 writes keys.json *atomically* (temp file ➜ fsync ➜ rename), so an interrupted save can never truncate your key file again. The atomic write needs a per-worker *directory* mount, so the upgrade includes a one-time key migration. On each node:
⚠️ *Stop your frontend workers FIRST*, before you migrate or upgrade. Nothing must be
Mike , thank you for documenting this. I’m sorry the setup and monitoring gaps cost you KAS and reduced your stake.
Your point about the dashboard is correct. “Active” reports the on-chain lifecycle state. It does not check whether Docker, kaspad, Traefik, the workers, or the attestor client are healthy and submitting. We need a separate operational status, a last-successful-attestation indicator, and an alert when submissions stop.
We have also recorded your feedback about recovery instructions, dependency checks, worker configuration, placeholder-address validation, and safe maintenance procedures.
Regarding the new 34.7 IGRA penalty: with 400,000 IGRA staked, one missed-attestation penalty is 34.72 IGRA under the current 1/11,520 ratio. That increase is consistent with one challenged missed duty. A reboot or Pause/Resume transaction does not add a penalty by itself. Zero Slashes means the contract recorded no fraud slash.
For attesting alone, one frontend worker with wallet 0 funded is enough. Five workers behind Traefik distribute the same request stream across several wallets. They do not multiply normal KAS consumption by five. The setup should also reject placeholder return addresses before a worker can send funds.
The lock period is global and shared by all mainnet attesters:
- Validation started at block 86,400 on 26 February 2026 at 13:07:35 UTC.
- The lock ends at block 15,638,400, projected for around 29 August 2026 at the present block rate.
- You can Pause/request exit before the lock ends. Once the transaction confirms and your status becomes PendingExit, missed-attestation penalty calls cannot succeed while you remain paused.
- The exit cooldown is 8,200 blocks, around 2 hours and 17 minutes.
- Once both the lock and cooldown have ended, claim your pending rewards and call Exit with your receiving address. The contract returns your effective stake after penalties and slashes.
For a temporary pause, bring the full stack up and confirm that it i
Hi IGRA Support — I’ve decided to cease all attesting going forward.
After experiencing repeated penalties, including another missed-attestation penalty when Traefik was paused, my total penalties have now reached approximately 382.21 IGRA. At this point, I’m no longer comfortable continuing to risk additional stake.
I will keep my IGRA node running to continue supporting the network, but I intend to safely pause/exit the attester and discontinue attestation duties permanently.
Please confirm the correct procedure for exiting without creating any additional missed-attestation penalties during the transition.