Nils Goroll

bugwash on the first point: We should improve the documentation: * mention overhead from the memory allocator * mention jemalloc + THP issues * provide links to alternative storage engines...

Please leave this ticket open until we improved the documentation. PRs welcome!

I'd like to briefly talk about this ticket again during bugwash: are we happy with the `onerror=` change from #3767 or do we want `alt=` also, or even something else...

VDD: Keep the `VSUB_closefrom()` change only, we should not need to fork once again. OK to merge.

Firstly, the intention behind this panic is a good one - to improve your system's uptime by restarting the varnish worker process if something goes terribly wrong. The respective parameter...

@Vormillion can you please let us know which binaries you are using? If you built varnishd yourself, please make the binary available for download somewhere. FWIW, the `PAN__DumpStruct+0x110b` offset looks...

> On a related issue, the stack trace points to a potential issue in our panic handler, but that is secondary and non-causal. `pan_pool()` is last in `pan_ic()` and looks...

I've spent some time with this today and noticed two things on the way, which are most likely not the root cause here: Regarding the garbled vxids, I first thought...

bugwash: it _could_ be an overrun because `vtc_varnish` uses `-l 2m`. TODO: check for overruns.

I've [reset `vsl_space` back to the 80m default](https://gitlab.com/uplex/varnish/libvdp-pesi/-/commit/671ec41f8904489b132efd9f7cd541b63ec016a5) and checked that the setting is applied. The problem persists. ``` *** v1 CLI RX 200 **** v1 CLI RX|vsl_space **** v1...