Question about scaling and race conditions

Question

Question about scaling and race conditions

rudolfbyker opened this issue 3 years ago · comments

Is running multiple Rendertron containers (e.g. AWS ECS) with a shared filesystem (e.g. AWS EFS) cache vulnerable to race conditions? Or is this supported?

Ravi Gidwani · Answer 1 · Sat May 15 2021 07:33:51 GMT+0800 (China Standard Time)

@rudolfbyker I am having similar setup and so far have not seen issues based on some internal testing only. This is yet to go to production, so can't comment much.
I would expect the EFS to be slower but have not benchmarked. Do share your findings too.

Rudolf Byker · Answer 2 · Sat May 15 2021 14:55:44 GMT+0800 (China Standard Time)

Thanks! I'm also Considering using the far gate "bind mount", which is faster, but ephemeral. But then the same question about race conditions needs to be answered. I will report back later if and when I know whether my setup works. :)

…

On Sat, 15 May 2021, 01:34 Ravi Gidwani, ***@***.***> wrote: @rudolfbyker <https://github.com/rudolfbyker> I am having similar setup and so far have not seen issues based on some internal issues only. This is yet to go to production, so can't comment much. I would expect the EFS to be slower but have not benchmarked. So share your findings. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#690 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACMPS7Q67CBNTDYIJ5SAGKLTNWXO3ANCNFSM44YYLU6Q> .

Ravi Gidwani · Answer 3 · Sat May 15 2021 15:11:39 GMT+0800 (China Standard Time)

Are you using rendertron on AWS fargate? I did read few articles where people setup puppeteer on AWS fargate and AWS lambda. But was wondering if rendertron performance will be lower on fargate compared to bare EC2 instances. On Fri, May 14, 2021, 11:55 PM Rudolf Byker ***@***.***> wrote:

…

Thanks! I'm also Considering using the far gate "bind mount", which is faster, but ephemeral. But then the same question about race conditions needs to be answered. I will report back later if and when I know whether my setup works. :) On Sat, 15 May 2021, 01:34 Ravi Gidwani, ***@***.***> wrote: > @rudolfbyker <https://github.com/rudolfbyker> I am having similar setup > and so far have not seen issues based on some internal issues only. This is > yet to go to production, so can't comment much. > I would expect the EFS to be slower but have not benchmarked. So share > your findings. > > — > You are receiving this because you were mentioned. > Reply to this email directly, view it on GitHub > < #690 (comment) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/ACMPS7Q67CBNTDYIJ5SAGKLTNWXO3ANCNFSM44YYLU6Q > > . > — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#690 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJSAUTCWBKICYEI7CM3NFTTNYLHZANCNFSM44YYLU6Q> .

Rudolf Byker · Answer 4 · Sat May 15 2021 19:16:45 GMT+0800 (China Standard Time)

Yes, I'm running Rendertron on AWS ECS with the "Fargate" launch type. Seems to work fine.

…

On Sat, 15 May 2021 at 09:11, Ravi Gidwani ***@***.***> wrote: Are you using rendertron on AWS fargate? I did read few articles where people setup puppeteer on AWS fargate and AWS lambda. But was wondering if rendertron performance will be lower on fargate compared to bare EC2 instances. On Fri, May 14, 2021, 11:55 PM Rudolf Byker ***@***.***> wrote: > Thanks! I'm also Considering using the far gate "bind mount", which is > faster, but ephemeral. But then the same question about race conditions > needs to be answered. I will report back later if and when I know whether > my setup works. :) > > On Sat, 15 May 2021, 01:34 Ravi Gidwani, ***@***.***> wrote: > > > @rudolfbyker <https://github.com/rudolfbyker> I am having similar setup > > and so far have not seen issues based on some internal issues only. This > is > > yet to go to production, so can't comment much. > > I would expect the EFS to be slower but have not benchmarked. So share > > your findings. > > > > — > > You are receiving this because you were mentioned. > > Reply to this email directly, view it on GitHub > > < > #690 (comment) > >, > > or unsubscribe > > < > https://github.com/notifications/unsubscribe-auth/ACMPS7Q67CBNTDYIJ5SAGKLTNWXO3ANCNFSM44YYLU6Q > > > > . > > > > — > You are receiving this because you commented. > Reply to this email directly, view it on GitHub > < #690 (comment) >, > or unsubscribe > < https://github.com/notifications/unsubscribe-auth/AAJSAUTCWBKICYEI7CM3NFTTNYLHZANCNFSM44YYLU6Q > > . > — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#690 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ACMPS7SWTNHC25UC7IS5UYDTNYNDTANCNFSM44YYLU6Q> .

Martin Splitt · Answer 5 · Thu Oct 06 2022 15:58:17 GMT+0800 (China Standard Time)

In general, as we're deprecating the project, you should look into alternative approaches to rendering on the web.