Snowflake bridge keeps triggering OOM killer

Seems like #40051 (closed) did not solve our problem. The server was down again this morning:

Jun 18 11:58:44 snowflake kernel: [240512.397993] sshd invoked oom-killer: gfp_mask=0x100cca(GFP_HIGH
USER_MOVABLE), order=0, oom_score_adj=0

[snip]

Jun 18 11:58:44 snowflake kernel: [240512.444250] [  pid  ]   uid  tgid total_vm      rss pgtables_bytes swapents oom_score_adj name
Jun 18 11:58:44 snowflake kernel: [240512.472145] [   1641]   106  1641   168150    93653  1089536        0             0 tor
Jun 18 11:58:44 snowflake kernel: [240512.473285] [   1642]   106  1642   548343   234687  2072576        0             0 snowflake-serve
Jun 18 11:58:44 snowflake kernel: [240512.476205] [   7901]   108  7901      621       21    40960        0             0 timeout
Jun 18 11:58:44 snowflake kernel: [240512.477206] [   7902]   108  7902   231410     2098   196608        0             0 proxy-go
Jun 18 11:58:44 snowflake kernel: [240512.477909] [  12745]   108 12745      621       22    45056        0             0 timeout
Jun 18 11:58:44 snowflake kernel: [240512.479216] [  12746]   108 12746   231410     3605   200704        0             0 proxy-go

[snip]

Jun 18 11:58:44 snowflake kernel: [240512.482848] oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/system.slice/system-tor.slice/tor@default.service,task=snowflake-serve,pid=1642,uid=106
Jun 18 11:58:44 snowflake kernel: [240512.485406] Out of memory: Killed process 1642 (snowflake-serve) total-vm:2193372kB, anon-rss:938748kB, file-rss:0kB, shmem-rss:0kB, UID:106 pgtables:2024kB oom_score_adj:0
Jun 18 11:58:44 snowflake kernel: [240512.496421] oom_reaper: reaped process 1642 (snowflake-serve), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB