== coolify-so coolify n=10 mem_med=320 MiB mem_max=371 MiB cpu_med=5.5% cpu_max=45.1% coolify-db n=10 mem_med=36 MiB mem_max=40 MiB cpu_med=0.6% cpu_max=2.4% coolify-realtime n=10 mem_med=49 MiB mem_max=49 MiB cpu_med=0.3% cpu_max=0.4% coolify-redis n=10 mem_med=6 MiB mem_max=9 MiB cpu_med=0.9% cpu_max=2.2% TOTAL-coolify 461 MiB (maior soma num mesmo instante) MAQUINA n=10 used_max=653 MiB available_min=6784 MiB (free -m: inclui dockerd/containerd/docker-proxy) == repouso-sem-canal coolify n=10 mem_med=114 MiB mem_max=118 MiB cpu_med=1.4% cpu_max=4.1% coolify-db n=10 mem_med=15 MiB mem_max=16 MiB cpu_med=0.4% cpu_max=2.3% coolify-proxy n=10 mem_med=36 MiB mem_max=36 MiB cpu_med=0.8% cpu_max=3.0% coolify-realtime n=10 mem_med=18 MiB mem_max=19 MiB cpu_med=0.3% cpu_max=0.3% coolify-redis n=10 mem_med=8 MiB mem_max=8 MiB cpu_med=0.7% cpu_max=2.2% coolify-sentinel n=10 mem_med=6 MiB mem_max=6 MiB cpu_med=0.1% cpu_max=0.6% deskcommcrm-app-1 n=10 mem_med=183 MiB mem_max=184 MiB cpu_med=0.1% cpu_max=1.1% deskcommcrm-redis-1 n=10 mem_med=1 MiB mem_max=1 MiB cpu_med=0.3% cpu_max=0.3% deskcommcrm-scheduler-1 n=10 mem_med=1 MiB mem_max=1 MiB cpu_med=0.0% cpu_max=0.2% deskcommcrm-srh-1 n=10 mem_med=7 MiB mem_max=7 MiB cpu_med=0.1% cpu_max=0.1% deskcommcrm-supabase-api-gw-1 n=10 mem_med=14 MiB mem_max=14 MiB cpu_med=0.5% cpu_max=1.9% deskcommcrm-supabase-auth-1 n=10 mem_med=12 MiB mem_max=12 MiB cpu_med=0.4% cpu_max=1.9% deskcommcrm-supabase-db-1 n=10 mem_med=216 MiB mem_max=219 MiB cpu_med=1.5% cpu_max=4.1% deskcommcrm-supabase-functions-1 n=10 mem_med=2 MiB mem_max=2 MiB cpu_med=0.4% cpu_max=2.0% deskcommcrm-supabase-imgproxy-1 n=10 mem_med=21 MiB mem_max=22 MiB cpu_med=1.1% cpu_max=4.6% deskcommcrm-supabase-meta-1 n=10 mem_med=38 MiB mem_max=39 MiB cpu_med=0.9% cpu_max=8.0% deskcommcrm-supabase-realtime-1 n=10 mem_med=50 MiB mem_max=50 MiB cpu_med=0.3% cpu_max=0.6% deskcommcrm-supabase-rest-1 n=10 mem_med=43 MiB mem_max=43 MiB cpu_med=1.0% cpu_max=3.1% deskcommcrm-supabase-storage-1 n=10 mem_med=29 MiB mem_max=30 MiB cpu_med=0.4% cpu_max=2.3% deskcommcrm-supabase-studio-1 n=10 mem_med=167 MiB mem_max=168 MiB cpu_med=0.7% cpu_max=7.0% deskcommcrm-supabase-supavisor-1 n=10 mem_med=32 MiB mem_max=32 MiB cpu_med=1.0% cpu_max=3.7% deskcommcrm-waha-1 n=10 mem_med=61 MiB mem_max=61 MiB cpu_med=0.1% cpu_max=0.1% deskcommcrm-worker-1 n=10 mem_med=110 MiB mem_max=110 MiB cpu_med=0.1% cpu_max=0.7% TOTAL-coolify 203 MiB (maior soma num mesmo instante) TOTAL-crm 364 MiB (maior soma num mesmo instante) TOTAL-supabase 629 MiB (maior soma num mesmo instante) MAQUINA n=10 used_max=1531 MiB available_min=5993 MiB (free -m: inclui dockerd/containerd/docker-proxy) == repouso-com-canal coolify n=10 mem_med=128 MiB mem_max=134 MiB cpu_med=0.8% cpu_max=3.4% coolify-db n=10 mem_med=17 MiB mem_max=18 MiB cpu_med=0.2% cpu_max=2.2% coolify-proxy n=10 mem_med=38 MiB mem_max=42 MiB cpu_med=0.2% cpu_max=0.3% coolify-realtime n=10 mem_med=16 MiB mem_max=19 MiB cpu_med=0.9% cpu_max=2.2% coolify-redis n=10 mem_med=8 MiB mem_max=9 MiB cpu_med=0.7% cpu_max=2.2% coolify-sentinel n=10 mem_med=7 MiB mem_max=7 MiB cpu_med=0.0% cpu_max=0.0% deskcommcrm-app-1 n=10 mem_med=202 MiB mem_max=212 MiB cpu_med=0.0% cpu_max=0.0% deskcommcrm-redis-1 n=10 mem_med=1 MiB mem_max=1 MiB cpu_med=0.5% cpu_max=2.2% deskcommcrm-scheduler-1 n=10 mem_med=2 MiB mem_max=3 MiB cpu_med=0.2% cpu_max=1.6% deskcommcrm-srh-1 n=10 mem_med=4 MiB mem_max=4 MiB cpu_med=0.1% cpu_max=0.1% deskcommcrm-supabase-api-gw-1 n=10 mem_med=15 MiB mem_max=15 MiB cpu_med=0.6% cpu_max=2.3% deskcommcrm-supabase-auth-1 n=10 mem_med=11 MiB mem_max=12 MiB cpu_med=0.3% cpu_max=1.6% deskcommcrm-supabase-db-1 n=10 mem_med=200 MiB mem_max=208 MiB cpu_med=1.9% cpu_max=4.5% deskcommcrm-supabase-functions-1 n=10 mem_med=3 MiB mem_max=3 MiB cpu_med=0.4% cpu_max=2.1% deskcommcrm-supabase-imgproxy-1 n=10 mem_med=17 MiB mem_max=19 MiB cpu_med=0.9% cpu_max=4.7% deskcommcrm-supabase-meta-1 n=10 mem_med=25 MiB mem_max=27 MiB cpu_med=1.8% cpu_max=7.4% deskcommcrm-supabase-realtime-1 n=10 mem_med=44 MiB mem_max=48 MiB cpu_med=0.4% cpu_max=0.8% deskcommcrm-supabase-rest-1 n=10 mem_med=44 MiB mem_max=44 MiB cpu_med=0.5% cpu_max=2.5% deskcommcrm-supabase-storage-1 n=10 mem_med=23 MiB mem_max=24 MiB cpu_med=0.5% cpu_max=1.9% deskcommcrm-supabase-studio-1 n=10 mem_med=104 MiB mem_max=106 MiB cpu_med=0.8% cpu_max=7.3% deskcommcrm-supabase-supavisor-1 n=10 mem_med=36 MiB mem_max=38 MiB cpu_med=0.3% cpu_max=0.6% deskcommcrm-waha-1 n=10 mem_med=103 MiB mem_max=104 MiB cpu_med=0.2% cpu_max=1.2% deskcommcrm-worker-1 n=10 mem_med=57 MiB mem_max=61 MiB cpu_med=0.4% cpu_max=1.2% TOTAL-coolify 225 MiB (maior soma num mesmo instante) TOTAL-crm 379 MiB (maior soma num mesmo instante) TOTAL-supabase 532 MiB (maior soma num mesmo instante) MAQUINA n=10 used_max=1360 MiB available_min=6164 MiB (free -m: inclui dockerd/containerd/docker-proxy) == conversa coolify n=48 mem_med=134 MiB mem_max=198 MiB cpu_med=6.1% cpu_max=105.1% coolify-db n=48 mem_med=18 MiB mem_max=21 MiB cpu_med=0.7% cpu_max=2.4% coolify-proxy n=48 mem_med=112 MiB mem_max=112 MiB cpu_med=0.8% cpu_max=2.6% coolify-realtime n=48 mem_med=21 MiB mem_max=21 MiB cpu_med=0.7% cpu_max=2.6% coolify-redis n=48 mem_med=8 MiB mem_max=9 MiB cpu_med=0.8% cpu_max=2.7% coolify-sentinel n=48 mem_med=8 MiB mem_max=8 MiB cpu_med=0.3% cpu_max=2.2% deskcommcrm-app-1 n=48 mem_med=206 MiB mem_max=210 MiB cpu_med=1.3% cpu_max=19.8% deskcommcrm-redis-1 n=48 mem_med=1 MiB mem_max=1 MiB cpu_med=0.3% cpu_max=0.3% deskcommcrm-scheduler-1 n=48 mem_med=2 MiB mem_max=3 MiB cpu_med=0.1% cpu_max=3.7% deskcommcrm-srh-1 n=48 mem_med=7 MiB mem_max=7 MiB cpu_med=0.0% cpu_max=0.1% deskcommcrm-supabase-api-gw-1 n=48 mem_med=16 MiB mem_max=20 MiB cpu_med=0.6% cpu_max=3.6% deskcommcrm-supabase-auth-1 n=48 mem_med=12 MiB mem_max=12 MiB cpu_med=0.3% cpu_max=1.9% deskcommcrm-supabase-db-1 n=48 mem_med=163 MiB mem_max=212 MiB cpu_med=1.3% cpu_max=7.8% deskcommcrm-supabase-functions-1 n=48 mem_med=3 MiB mem_max=3 MiB cpu_med=0.4% cpu_max=2.1% deskcommcrm-supabase-imgproxy-1 n=48 mem_med=24 MiB mem_max=28 MiB cpu_med=1.0% cpu_max=5.7% deskcommcrm-supabase-meta-1 n=48 mem_med=29 MiB mem_max=38 MiB cpu_med=1.8% cpu_max=9.4% deskcommcrm-supabase-realtime-1 n=48 mem_med=51 MiB mem_max=51 MiB cpu_med=0.4% cpu_max=2.7% deskcommcrm-supabase-rest-1 n=48 mem_med=44 MiB mem_max=46 MiB cpu_med=1.2% cpu_max=5.8% deskcommcrm-supabase-storage-1 n=48 mem_med=24 MiB mem_max=25 MiB cpu_med=0.3% cpu_max=2.1% deskcommcrm-supabase-studio-1 n=48 mem_med=117 MiB mem_max=125 MiB cpu_med=1.6% cpu_max=8.4% deskcommcrm-supabase-supavisor-1 n=48 mem_med=40 MiB mem_max=40 MiB cpu_med=0.8% cpu_max=3.1% deskcommcrm-waha-1 n=48 mem_med=108 MiB mem_max=108 MiB cpu_med=0.1% cpu_max=0.4% deskcommcrm-worker-1 n=48 mem_med=117 MiB mem_max=118 MiB cpu_med=0.3% cpu_max=1.9% TOTAL-coolify 368 MiB (maior soma num mesmo instante) TOTAL-crm 446 MiB (maior soma num mesmo instante) TOTAL-supabase 569 MiB (maior soma num mesmo instante) MAQUINA n=48 used_max=1688 MiB available_min=5836 MiB (free -m: inclui dockerd/containerd/docker-proxy) == F4 Passo 3 (regra de dimensionamento, 2026-10-07; numeros acima, LC_ALL=C) Janelas: F1 15:31:00-15:35:52Z, F2 15:45:07-15:49:58Z (sessao WAHA em FAILED: o QR expirou antes da janela), F3 16:05:51-16:12:30Z. Todas as janelas rodaram DENTRO do flock /root/.vps-teste.lock; TOTAL-terceiros = 0 em todas as amostras (nenhuma linha de grupo terceiros no docker stats). DIVERGENCIA (o plano supoe "sem swap"): desde 2026-10-07 13:46:10Z existe /swapfile de 4 GiB (persistente no fstab), criado por OUTRA sessao, depois de 14 OOM do kernel em processos node de terceiros (13:32-13:44Z, antes da rodada 1 do PR 4). Durante as janelas o swap tinha ~1,6 GiB em uso, e ~1611 MiB eram paginas dos NOSSOS contêineres (16:26:59Z: coolify 280, supabase 797, crm 534; ver swap-e-oom.txt). O MEM USAGE do docker stats e o "used" do free -m NAO contam swap: os numeros abaixo sao RAM residente numa maquina COM swap. Regua 1 (a do plano, used da MAQUINA na conversa, amostras com TOTAL-terceiros=0): minimo = 1688 MiB. Regua 2 (soma no mesmo instante dos tres grupos nossos, conversa): 1362 MiB (coolify 368, supabase 569, crm 446 sao maximos por grupo, nao no mesmo instante). Regua 3 (residente + swap dos nossos, limite superior, swap lido 15 min depois da janela): 1688 + 1611 = 3299 MiB. Folga ate os mem_limit do CRM (conversa): app 768-210 + worker 512-118 + waha 1280-108 = 2124 MiB. memory.peak (com cache, desde a subida): supabase 1828, coolify 1019 MiB; folga ate o peak: supabase 1828-569=1259, coolify 1019-368=651. Recomendado, leitura A (minimo + folga CRM + folga ate o peak): 1688 + 2124 + 1910 = 5722 MiB (5,6 GiB) -> <= 6,2 GiB -> 8 GB. Recomendado, leitura B (minimo + folga CRM + peak inteiro): 1688 + 2124 + 2847 = 6659 MiB (6,5 GiB) -> > 6,2 GiB -> 16 GB. Recomendado, leitura A sobre a regua 3: 3299 + 2124 + 1910 = 7333 MiB (7,2 GiB) -> 16 GB. DECISAO 8 GB x 16 GB: NAO TOMADA pelo executor. Depende (1) de qual leitura da regra vale e (2) de remedir sem swap ou aceitar a regua 3. Maquina inteira sem swap: NAO MEDIDA. VPS de 4 GB: NAO MEDIDA. Disco: imagens 19,84 GB no Docker (TODAS, inclui Coolify e terceiros), banco+anexos do Supabase 105 MiB (volumes), / com 37G usados de 196G. OOM do kernel durante as janelas da F: 0 (os 14 sao de 13:32-13:44Z). Conteineres nossos com OOMKilled/reinicio: 0 nas tres janelas.