Hi! I'm using a hosted OpenClaw instance on the Kilocode dashboard. I recently clicked redeploy/start, and my bot woke up with a completely blank memory (zero history files mounted in the /root environment). The underlying Fly.io app name is acct-2199959cffbc2d8620ed. It looks like my instance was provisioned with a new empty storage volume instead of reattaching my original persistent volume. Can someone please check my account and manually remap my original memory volume back to my current machine setup so I can recover the bot's memory?
#Redeployed hosted instance woke up with wiped memory (wrong volume mounted)
1 messages · Page 1 of 1 (latest)
I can’t check Kilocode accounts or remap Fly volumes from here (no access to their dashboard / your Fly org), but this is very likely recoverable because Fly volumes don’t get deleted on redeploy unless someone explicitly destroys them.
A few concrete things to verify:
- Confirm where OpenClaw expects its persisted state
- OpenClaw keeps sessions + history under
$OPENCLAW_STATE_DIR(default is~/.openclaw, but on Fly it’s commonly set to/data). - Docs: https://docs.openclaw.ai/help/faq#where-does-openclaw-store-its-data
- Fly guide (shows the intended
/datamount +OPENCLAW_STATE_DIR=/data): https://docs.openclaw.ai/install/fly
- Check whether the machine is running with the “wrong” empty volume
If you have Fly access, these are read-only and safe to paste here:
fly machines list -a acct-2199959cffbc2d8620ed
fly volumes list -a acct-2199959cffbc2d8620ed
What we’re looking for:
- multiple volumes (old one with your real data + new empty one)
- a region mismatch (volumes are region-bound; a new machine in a different region can’t attach the old volume)
- Quick on-box sanity check (also safe)
fly ssh console -a acct-2199959cffbc2d8620ed
echo "OPENCLAW_STATE_DIR=$OPENCLAW_STATE_DIR"
ls -la /data || true
ls -la /data/agents || true
ls -la /root/.openclaw || true
openclaw --version || true
If it is the wrong volume
Recovery on Fly is generally: attach the original volume to the running machine (mounted at /data) and ensure OPENCLAW_STATE_DIR=/data. If the old volume is in a different region, you usually need to bring up a machine in that region (or restore from a snapshot).
If you don’t have Fly access (Kilocode-managed)
Then the action item is: open a Kilocode support ticket and ask them to:
- list volumes for
acct-2199959cffbc2d8620ed - identify the original volume ID (by created time/size) and reattach it to the active machine
- confirm mount destination is
/dataand envOPENCLAW_STATE_DIR=/data
If you paste the machines list + volumes list output (redacting anything sensitive), I can tell you exactly what to ask them to reattach (and whether it’s just a region mismatch).
Thanks for the detailed reply! Yes, this is a Kilocode-managed instance, so I don't have direct Fly CLI access to run those commands, and my bot confirmed it can't run them from inside its container either. It also noted that right now OPENCLAW_STATE_DIR doesn't seem to be mapped to /data and is just defaulting to /root/.openclaw on a fresh 1.4MB volume.
Since Kilocode manages this, what is the best and fastest way for me to open a support ticket with your team? I just need an admin to check the volumes for acct-2199959cffbc2d8620ed, find my original persisted volume, and reattach it to the active machine so my bot gets its brain back!
We’re not the Kilocode/Kilo Claw ops team in this Discord (so nobody here can access your Fly app or remap volumes), but the fastest path is to go straight to Kilo support and give them the exact “reattach volume + fix mount/env” request.
Fastest way to open a Kilocode/Kilo ticket
Use Kilo’s support page and submit the contact form:
https://kilo.ai/support
(That page also mentions a “Discord Support” option, but the web form is usually the cleanest for account-specific infra work like volume reattachment.)
What to put in the ticket (copy/paste template)
Subject: Kilo Claw / OpenClaw state volume not reattached after redeploy (app: acct-2199959cffbc2d8620ed)
Body:
- Fly app:
acct-2199959cffbc2d8620ed - Action taken: clicked redeploy/start in Kilocode dashboard
- Symptom: bot came up with blank sessions/memory; state is now under
/root/.openclawon a fresh tiny volume (~1.4MB) - Expected: persistent volume mounted at
/datawithOPENCLAW_STATE_DIR=/data(so sessions + workspace persist) - Request: please locate my original persisted Fly volume for this app and reattach it to the active machine (mount destination
/data), and restoreOPENCLAW_STATE_DIR=/data(or whatever your standard is), then restart the machine.