https://discord.cloudflare.com logo
Join Discord
Powered by
# workers-discussions
  • i
  • w
    Basically how the server works now is that it reads the files once, into memory and then clients send requests for certain sets of files and the server sends back those files from memory, it never needs to write to disk either. So looks like this might need a push CDN then? By the way, thanks for helping out with these questions, I appreciate it.
  • i
    It doesn't need to be push from the needs I see. Two alternatives depeding on type of file:
  • i
    1. if the files are all the same for everyone, just different sets of them I would use a mix of this (https://developers.cloudflare.com/workers/examples/aggregate-requests) and either store the files in the cache once they are there in the Worker (either grouped together or by file) or in KV. There was an example of this second process, can't find it right now.
  • i
    KV prevents a read forever (unless a TTL is stated) if you put each file into it once it's fetched from the server (you could also do that proactively). Size is max 25MB per file or you need to shard them.
  • i
    Cache is local per POP, but free. It will expire. Limits in size here as well, see link above.
  • i
    2. Use the normal CF cache, saving each file set when outbound with a decent config of 
    cache-control
    headers. This won't be the same as the above in speed and reduced load on the server, but it's cheaper if the server runs regardless.
  • i
    ---
  • i
    I would personally use the first and possibly move the logic of the grouping into the Worker. You could then store the files in something like S3 or directly into KV, removing the server completely.
  • i
    Might not be possible depending on your architecture and needs.
  • w
    Hm the first way looks like it could work for me actually. I'm going to look into how I can move my currently server logic into a Worker. Thanks again for your help.
  • i
    No worries! Just ask if you need anything 🙂
  • a

    https://cdn.adine.tech/chrome_XuLm1Zugwf.pngâ–¾

  • a
    So i got help from cloudflare community
  • a
    Great
  • s
    is there hard limit for parallel read for workers kv?
  • c
    By a single worker or multiple workers? For all workers I don't believe theres an account wide rate limit on KV reads (but there is for writes). For an individual worker, I'd also be curious if there is a rate limit on KV reads for a single request event.
  • s
    for single worker
  • y
  • s
    Is the limit 1000ops per invocation?
  • e
    Check "subrequests" section of the limits docs, it does answer both total requests per invocation and simultaneous open connections tldr - 50 subrequests (that might apply only to fetch, didnt run into KV limits myself) - 6 open connections, all (fetch, kv, cache) counts towards limit
  • d
    "Remember me" checkbox stopped working on a login page to CF dashboard. Anyone?
  • m
    Yeah, I have the same issue.
  • i
    I believe this is best reported in the Community (https://community.cloudflare.com), but it happens sometimes to me as well.
  • t
    Anyone doing CI/CD with workers? I've seen https://blog.cloudflare.com/a-ci/ w/ Travis CI. I've been looking at GoCD and considering using that, it seems simpler/more modern than Jenkins, and still free.
  • g
  • t
    not using Github sadly, but that's cool!
  • e
    maybe its about time to add workers to https://www.waypointproject.io/
  • g
    The action is basically just installing wrangler and running 
    wrangler publish
    so it should be pretty easy to build something equivalent in any other CI/CD platform.
  • g
    There's also a serverless integration (https://www.serverless.com/framework/docs/providers/cloudflare/guide/intro/), but I think that's no longer actively developed (possibly deprecated?)
1...616263...2509Latest