I get the following errors when running pvagw:
double free or corruptionmalloc(): unaligned tcache chunkBus error
Logs also show:
stale GET_FIELD, bad request id, connection resets
My current setup:
- ~100 IOCs (~15k PVs each → ~1500k total)
- Many heavy AD PVs (NDArray, image, ArrayData)
- Single VM gateway, ~5 client machines
- Initially one process, now split across multiple processes
What I tried:
- Splitting load (15-20 IOCs per process) → improves but still crashes
- Blocking NDArray PVs → more stable
- Tuning:
channel_cache (lower helps)getholdoff (0.05–0.1 helps, large values crash faster)
The issue becomes reproducible when scaling clients:
- 3 clients → ~2.7 GB/s → stable
- 4 clients → ~3.4–4 GB/s → gateway crashes
At this point, these requests are almost equal to 400k*3 PVs at a time to gateway server.
Is gateway able to handle this amount of traffic?
Observation: Crashes correlate with sustained load, rising threads, and heavy GET/monitor traffic.
I get the following errors when running pvagw:
double free or corruptionmalloc(): unaligned tcache chunkBus errorLogs also show:
stale GET_FIELD,bad request id, connection resetsMy current setup:
What I tried:
channel_cache(lower helps)getholdoff(0.05–0.1 helps, large values crash faster)The issue becomes reproducible when scaling clients:
At this point, these requests are almost equal to 400k*3 PVs at a time to gateway server.
Is gateway able to handle this amount of traffic?
Observation: Crashes correlate with sustained load, rising threads, and heavy GET/monitor traffic.