Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Taaalk [Update: site now back up]

taaalk.co

Post reply on HN
Print view View original

31–40 of 60 posts

Re: Taaalk [Update: site now back up]

#31

Post by viraptor »

satvikpendem wrote: post #14

Ah, the ol' HN hug of death. For future reference (and I haven't been able to access the site yet, so this may or may not be relevant to Taaalk), if the site has a landing page, make it with static files, and put it behind a CDN or just use a free service like Netlify. Put the actual app behind a subdomain which you can then host however you want. This way, people who visit can still check out what the service does a…

With less architecture work: enable local caching of the front page/resources on your nginx/apache/whatever. They should be able to ride the HN wave on almost any hardware.

Re: Taaalk [Update: site now back up]

#32

Post by Bnshsysjab »

FailMore wrote: post #5

Is the site very slow for everyone? Normally it's very fast for me but suddenly it's slow.

Once this is all over I would love to see what kind of metrics you have of hits per minute average vs HN, I’m super curious how much traffic sites get from here.

Good luck with the remediations :)

Re: Taaalk [Update: site now back up]

#33

Post by KittyMagician »

FailMore wrote: post #23
jtchang wrote: post #18

I would post your nginx.conf

Hey, thanks for writing to me. I'm using Hetzner Cloud. They have a 'Rescale' option which I used. To activate it I had to turn off my server. I did that. Then rescaled and it went on again. When I log into my server or run 'git push dokku master' everything seems the same/is up to date. But getting a 404 for the entire app. Server logs seem to be ok (found with dokku logs taaalk -t). So I'm just very confused and ve…

Hey FailMore. I've dropped you an email to provide some tips that may help.

Re: Taaalk [Update: site now back up]

#34

Post by gtirloni »

harpratap wrote: post #29
FailMore wrote: post #15

Earlier quoted context omitted.

Hey, I tried to scale up my server mid hn ride and now I'm just getting a 404. I'm not experienced with doing this, and I can log into the server and everything seems to be working/up to date. So I don't understand why it's not working. I'm using a Hetzner, and I'm bummed :(

This is why standardized solutions like K8s are necessary (or at least a managed container platform). Right now we have no idea about your stack or how it is hosted. With containers it wouldn't have mattered. And with K8s it would have been a simple `kubectl scale deploy/taaalk --replicas 10` regardless of on-prem, cloud X, Y or Z.

> regardless of on-prem, cloud X, Y or Z

Are you speaking from experience? Because my experience with k8s is far from that.

Re: Taaalk [Update: site now back up]

#36

Post by FailMore »

KittyMagician wrote: post #33
FailMore wrote: post #23

Earlier quoted context omitted.

Hey, thanks for writing to me. I'm using Hetzner Cloud. They have a 'Rescale' option which I used. To activate it I had to turn off my server. I did that. Then rescaled and it went on again. When I log into my server or run 'git push dokku master' everything seems the same/is up to date. But getting a 404 for the entire app. Server logs seem to be ok (found with dokku logs taaalk -t). So I'm just very confused and ve…

Hey FailMore. I've dropped you an email to provide some tips that may help.

Thank you, I think I just got it. Useful for the future/my education!

Re: Taaalk [Update: site now back up]

#37

Post by harpratap »

gtirloni wrote: post #34
harpratap wrote: post #29

Earlier quoted context omitted.

This is why standardized solutions like K8s are necessary (or at least a managed container platform). Right now we have no idea about your stack or how it is hosted. With containers it wouldn't have mattered. And with K8s it would have been a simple `kubectl scale deploy/taaalk --replicas 10` regardless of on-prem, cloud X, Y or Z.

> regardless of on-prem, cloud X, Y or Z Are you speaking from experience? Because my experience with k8s is far from that.

Yes. Our on-prem works a lot like GKE, for users the difference is negligible.

Re: Taaalk [Update: site now back up]

#39

Post by boyter »

viraptor wrote: post #31
satvikpendem wrote: post #14

Ah, the ol' HN hug of death. For future reference (and I haven't been able to access the site yet, so this may or may not be relevant to Taaalk), if the site has a landing page, make it with static files, and put it behind a CDN or just use a free service like Netlify. Put the actual app behind a subdomain which you can then host however you want. This way, people who visit can still check out what the service does a…

With less architecture work: enable local caching of the front page/resources on your nginx/apache/whatever. They should be able to ride the HN wave on almost any hardware.

For static websites you don't even need that. The OS disk cache kicks in and it becomes more about bandwidth than disk performance.

Re: Taaalk [Update: site now back up]

#40

Post by bostonsre »

harpratap wrote: post #29
FailMore wrote: post #15

Earlier quoted context omitted.

Hey, I tried to scale up my server mid hn ride and now I'm just getting a 404. I'm not experienced with doing this, and I can log into the server and everything seems to be working/up to date. So I don't understand why it's not working. I'm using a Hetzner, and I'm bummed :(

This is why standardized solutions like K8s are necessary (or at least a managed container platform). Right now we have no idea about your stack or how it is hosted. With containers it wouldn't have mattered. And with K8s it would have been a simple `kubectl scale deploy/taaalk --replicas 10` regardless of on-prem, cloud X, Y or Z.

The guy has a single web app. Not sure using kubernetes would make it easier.
Post reply on HN