We've been using Twingate for about six months to secure access to our internal dev and analytics tools. Overall, it's been solid, but we've hit a weird snag recently that I can't pin down.
Our team is distributed across US, EU, and APAC. Users in APAC (specifically Singapore and Sydney) are reporting intermittent "resource not found" errors when trying to access certain resources, but *only* during their local business hours (roughly 9 AM to 5 PM local). Our EU and US colleagues can access the same resources without issue at that same moment in UTC. The resources themselves are definitely online and reachable.
Here's what we've checked so far:
* Twingate Connectors show healthy status across all regions.
* No patterns in the Twingate logs that correlate with the outages—just the user-side error.
* User client apps (macOS and Windows) are up-to-date.
* The resources are assigned to all relevant user groups correctly.
It feels like a caching, DNS, or time-synchronization issue maybe tied to a specific relay? Has anyone else experienced timezone-specific or geographically-correlated access problems? I'm starting to look at network policy schedules, but we haven't configured any.
My next step is to open a support ticket, but I wanted to see if the community had ideas on specific logs to check or connector settings we might have missed.
Time-based network policies would've been my first guess, but you're already looking there. That "intermittent during business hours" screams scheduled scaling to me, not access.
Are those resources in APAC regions on AWS or Azure? If so, check if they're on a dev/test schedule that spins down instances overnight local time - but your Sydney colleague's 9 AM might still be your cloud region's previous day. A stopped instance would give a "not found" from the Twingate connector.
You could run a cheap script to poll the resource endpoint from different regions every 5 minutes for a day. CloudWatch Synthetics or just a cron job on a cheap VM. Map the failures against your cloud provider's billing dashboard - sometimes the cost timeline shows the shutdowns.
- elle
That's a really tricky one. That "only during local business hours" pattern is super weird. You mentioned the resources are online, but have you verified they're responding from an APAC network path at that exact time? Could be a geo-IP rule or load balancer config misbehaving.
Since it's only the user-side error in the logs, maybe run a continuous trace route from a Sydney machine to the resource during their next outage window? Might spot a hop that only exists during business hours.
What would you recommend as the next diagnostic step?