Our Linux dedicated server (hosted Docker/Pterodactyl container, 63 -mod entries + 1 -serverMod) crashes when a client that is missing most of the server's mods tries to join. The server writes

Player Unknown (<dpnid>) kicked from server: 115 (Client is missing a mod which is on the server. (XZone STALKER Mutants; LIGHTS; TrashBin; ...

and the process dies in the same second: crash_*.log says "Unhandled exception / Reason: Unknown" and a minidump is written. The host then restarts the server automatically, so one join attempt by someone with the wrong mod set means a restart for everyone online.

• Kicks with a short list (1-2 missing mods, ~150 characters) never crash.

• One kick with a message of about 1015 characters did not crash. Every longer one did (6 crashes so far).

• This looks like a fixed buffer of about 1 KB for the kick reason text.

Seen on 1.29.163709 (2 crashes), 1.30.164078 (1) and 1.30.164200 (3: 22.10, 26.10 09:52 and 26.10 19:29 UTC).

Minidump analysis (attached dumps, 1.30.164200.56)

• Signal 6 (SIGABRT), raised by glibc (crashing thread RIP at libc.so.6+0x8aeec).

• Return addresses on the crashing thread's stack: DayZServer+0xa2a702 -> libc.so.6+0x116602 -> +0x116d20 -> +0x118212 -> abort. That looks like the FORTIFYSOURCE __chk_fail path ("buffer overflow detected").

• On 1.30.164078 the same chain starts at DayZServer+0xa1c922.

• Offsets are relative to the DayZServer image base 0x400000.

• The kick text "Client is missing a mod which is on the server. (XZone STALKER Mutants; LIGHTS; TrashBin; MBM_ApocalypseTruck; ..." is in the dump memory.

1. Run the Linux DayZServer 1.30.164200 with enough mods that their names joined with "; " are longer than ~1 KB. We have 63 mods.

2. Connect with a client that is missing most of those mods.

3. The server logs kick 115 with the full list of missing mods and aborts.

the client is kicked (the message cut short if needed) and the server keeps running.