Open howardjohn opened 1 month ago
I have a sneaking suspicion this is related to #34334; notice my comment. I need to refresh myself on the flow, but that debug assert is probably failing because of a poor assumption
Thanks for reporting this. I'll follow up on @keithmattix's comment next week (should have more time to dig into this).
in isInitialRequestForLegacyWildcard
this time. There is a core dump.
Note: this was originally reported privately
Description: Crash logs and another with the same issue Version: Envoy f4930b1ffd7c7bffec32e73f4b7c8482a72dc23a (but Istio build). From yesterday. Core dump
The assertion has been there for a year. We do not always run these tests, so it's likely not a recent regression in Envoy.
Unfortunately, our control plane logs were truncated so we do not have the exact XDS info going on at the time. Probably I can run the test a few more times and get it if it helps. I do know that at the time of the trash, the last XDS push should have been approx 15s prior.
cc @keithmattix