Partial outage in FRA
The network is back up. We're continuing to monitor.
1 earlier update
One of the VM hosts has lost networking. We're investigating.
All systems operational
Real-time and historical status for exe.dev VMs, SSH, API, dashboard, and networking.
Past 90 daysPast 30 days
- No incidents
- Degradation
- Partial outage
- Full outage
- Maintenance
-
VMs99.81% uptime
4 components
-
Running VMsNo recorded downtime in the past 90 days
-
VM creation and lifecycle99.81% uptime
-
SSH to VMs99.81% uptime
-
HTTPS to VMsNo recorded downtime in the past 90 days
-
-
CLI and API99.98% uptime
2 components
-
exe.dev CLI99.98% uptime
-
API and MCP server99.98% uptime
-
-
Dashboard99.98% uptime
5 components
-
Dashboard99.98% uptime
-
Login and signup99.98% uptime
-
Billing and checkoutNo recorded downtime in the past 90 days
-
IntegrationsNo recorded downtime in the past 90 days
-
Usage metricsNo recorded downtime in the past 90 days
-
-
Shelley99.98% uptime
2 components
-
LLM gatewayNo recorded downtime in the past 90 days
-
Shelley agent99.98% uptime
-
How to read this page
We publish host-level and regional events as well as platform-wide ones. Each incident lists the regions and components it affected, and most affect a subset of VMs in one region. An incident listing no components affected all of them; one listing only a region counts against the VM components, which are the only regional ones.
Each row covers the last 90 days. A day takes the color of the worst incident that touched any of the row's components: full outage, then partial outage, degradation, maintenance.
Uptime is the share of those days with no outage open on any component in the row. A full or partial outage counts as downtime from the moment it was declared until it was first mitigated or resolved, and a partial outage counts the same as a full one. Overlapping incidents count once. Degradation and maintenance are shown but are not downtime. Incidents caused by an upstream provider are shown and marked, but are not counted. Percentages are rounded down.
Durability is separate from availability: a VM's disk streams its changes to a backup on another server in the same regional cluster, a few seconds behind, so the VM can be restarted there if its server fails. See the architecture page.
Regions
- Dallas, USA
- Frankfurt, Germany
- London, UK
- Los Angeles, USA
- New York, USA
- Oregon, USA
- Singapore
- Sydney, Australia
- Tokyo, Japan
- Virginia, USA
Past incidents
VM Operations Failing
Network traffic has resumed and VM operations in LAX are now working.
2 earlier updates
We've identified the issue, applied a fix and are now monitoring connections.
We're investigating issues creating and connecting to VMs in our LAX region
Some VMs down on one host
We have migrated away from this host.
1 earlier update
One of our infrastructure hosts rebooted. The VMs on that host are back online. We are in the process of moving them to other hosts while we determine the cause.
LLM Gateway Elevated Errors
We had a bad rollout which resulted in some LLM requests failing, this change has been rolled back.
1 earlier update
We're investigating elevated errors in the LLM gateway.
Access to exe.dev was down for about 12 minutes, interrupting SSH, some Shelley and API access, and some website access
Due to a broken update, access to exe.dev was lost for about 12 minutes. This interrupted ssh to exe.dev, some Shelley and API access, and some website access.
Some VMs are having disk problems
We've fixed the underlying issues. Please e-mail support@exe.dev if need be.
2 earlier updates
A fix has been implemented to skip disk upgrades if errors occur. We are currently repairing all VMs impacted.
A limited number of VMs are showing disk errors due to a broken upgrade. We are actively working to resolve the problem. At this point, if your VM isn't working, please run "ssh exe.dev restart <vm-name>" (or find the restart button in the VM detail page in the web UI). If that doesn't work, please e-mail support@exe.dev with the VM name.
GitHub Integration Connectivity
Resolved
5 earlier updates
We are currently monitoring the fix.
A fix is currently being rolled out
A fix has been implemented and will be deployed with the next release.
We've identified an issue where a stale attached GitHub repository can cause github.int.exe.xyz clones for otherwise valid repos to fail. Users can re-grant or detach the stale GitHub repo integration while we prepare a fix
We're investigating an issue with some VMs not being able to communicate or refresh tokens from GitHub
Connectivity Issues in FRA
One of our upstream providers entered a small maintenance window. During this time, customers may have experienced brief connectivity issues.
Past maintenance
Maintenance: Server Migrations and Upgrades
Finished.
1 earlier update
We're actively migrating to new hardware across all regions.
