Hi specialone-coder,

For testing, I deployed a new project using the template.

The error message only appears the first time.

After redeploying the Postiz service while all other services are up and running,

there are no errors. I was able to register an account and got redirected to the calendar page. (Check the screenshots attached.)

Hi, thank you so much for testing this!

Glad it works on a fresh deploy. To clarify the exact scenario where I hit the issue:

After the initial setup and first successful redeploy (where everything works), if I redeploy the Postiz service again, whether to add new environment variables (like TikTok credentials) or even with absolutely no changes at all, the backend dies again with the same connection refused error on port 3000.

So the pattern is:

1. Fresh deploy → race condition → redeploy Postiz → works

2. Redeploy Postiz again for any reason → breaks again

It seems like the race condition with Temporal is not a one-time thing but happens on every subsequent redeploy of the Postiz service. Have you been able to reproduce this by redeploying the Postiz service a second or third time after it’s fully stable?​​​​​​​​​​​​​​​​

Solved it by adding this script in the postiz service Settings > Deploy > Custom Start Command

sh -lc "(node -e 'const net=require(\"net\");const sleep=ms=>new Promise(r=>setTimeout(r,ms));const parse=(v,d)=>{if(!v)return null;try{if(v.includes(\"://\")){const u=new URL(v);return{host:u.hostname,port:Number(u.port||d),name:v}}}catch{}const m=v.match(/^\\[?(.*?)\\]?:([0-9]+)$/);return m?{host:m[1],port:Number(m[2]),name:v}:null};const probe=t=>new Promise((res,rej)=>{const s=net.connect({host:t.host,port:t.port});s.setTimeout(2000);s.on(\"connect\",()=>{s.destroy();res()});s.on(\"timeout\",()=>{s.destroy();rej(new Error(\"timeout\"))});s.on(\"error\",e=>{s.destroy();rej(e)})});async function wait(t,timeout=300000){const end=Date.now()+timeout;let n=0;while(Date.now()<end){n++;try{await probe(t);console.log(\"ready:\",t.name);return}catch(e){const b=Math.min(1000*Math.pow(1.5,n),10000);console.log(\"waiting:\",t.name,e.message,\"retry in\",Math.round(b),\"ms\");await sleep(b)}}console.error(\"timeout waiting for\",t.name);process.exit(1)};(async()=>{for(const t of [parse(process.env.TEMPORAL_ADDRESS,7233),parse(process.env.DATABASE_URL,5432),parse(process.env.REDIS_URL,6379)].filter(Boolean))await wait(t)})().catch(()=>process.exit(1))') && nginx && (pnpm run pm2-run & PM2PID=$!; node -e 'const net=require(\"net\");const sleep=ms=>new Promise(r=>setTimeout(r,ms));const probe=()=>new Promise((res,rej)=>{const s=net.connect({host:\"127.0.0.1\",port:3000});s.setTimeout(2000);s.on(\"connect\",()=>{s.destroy();res()});s.on(\"timeout\",()=>{s.destroy();rej(new Error(\"timeout\"))});s.on(\"error\",e=>{s.destroy();rej(e)})});(async()=>{const end=Date.now()+180000;let n=0;while(Date.now()<end){n++;try{await probe();console.log(\"backend port ready: 3000\");break}catch(e){const b=Math.min(1000*Math.pow(1.4,n),5000);console.log(\"waiting for backend 3000:\",e.message,\"retry in\",Math.round(b),\"ms\");await sleep(b)}}if(Date.now()>=end){console.error(\"backend never became ready on 3000\");process.exit(1)}let fails=0;for(;;){await sleep(5000);try{await probe();fails=0}catch(e){fails++;console.log(\"backend 3000 healthcheck failed\",fails,e.message);if(fails>=3){console.error(\"backend 3000 unhealthy, exiting container\");process.exit(1)}}}})().catch(()=>process.exit(1))' || { kill $PM2PID 2>/dev/null || true; exit 1; }; wait $PM2PID)"

Great to hear you've fixed the issue. Consider testing your script to see if it solves it better. Thanks.