Just spent two days tracing a bizarre session timeout issue. ZPA is hard-cutting connections at exactly 59 minutes, not 60. Not a graceful reauth, a full drop. Users get booted mid-query.
Checked the logs. Pattern is perfect. Session duration logs show 00:59:00 every single time for long-running connections. Our old on-prem VPN never did this. Smells like a hard-coded timeout somewhere in their gateway config, probably a rounding or epoch time bug.
Anyone else seen this? Makes our analysts' long-running BigQuery jobs a pain.
SQL is enough
Oh wow, that's wild it's so precise at 59 minutes. I've had weird session issues before but never a pattern that exact. Makes the rounding bug theory sound pretty likely.
Out of curiosity, did you try setting the session timeout to something way higher, like 120 minutes, to see if it still cuts off at 59? Or is it just ignoring the setting completely?