How to use multiple S3 buckets

Hello, I’m trying to connect a second S3 bucket to my Node.js companion server.

After looking at the configuration, I did not find any option to specify more than one bucket for the same instance, so I thought about adding a second instance on another route.

So far so good, but as soon as I start the server, I get this error:

TypeError: Cannot add property 1, object is not extensible
    at Array.push (<anonymous>)
    at ..\node_modules\@uppy\companion\lib\server\logger.js:14:22
    at Array.forEach (<anonymous>)
    at Object.exports.setMaskables (..\node_modules\@uppy\companion\lib\server\logger.js:13:15)
    at maskLogger (..\node_modules\@uppy\companion\lib\companion.js:181:12)
    at Object.module.exports.app (..\node_modules\@uppy\companion\lib\companion.js:61:5)
    at createUppyCompanion (..\src\routes\uppy\createUppyEndpoint.ts:43:20)
    at Object.createUppyEndpoint (..\src\routes\uppy\createUppyEndpoint.ts:16:14)
    at Object.<anonymous> (..\src\routes\uppy\uppyArchiveRoute.ts:4:33)
    at Module._compile (internal/modules/cjs/loader.js:1085:14)

I can temporary fix this error by commenting out this line:

exports.setMaskables = (maskables) => {
    maskables.forEach((i) => {
        valuesToMask.push(escapeStringRegexp(i));
    });
    //Object.freeze(valuesToMask);
};

Now, my question is, if this is a bug or how to support multiple buckets otherwise?

[mifi]

I think just removing object.freeze is not the right way to do it, because then two app instances created with different configs will compete to overwrite this singleton variable, as well as cause issues with the other singletons/globals that are created. I think it boils down to whether we want to do the effort of rewriting all the singleton global state and instead place it inside the companion instance / closure.

if not ,then I think we should throw a better error message if someone tries to initialize the companion app twice.

[baba43]

Cool. Out of curiosity, may you share what is your use case for uploading to multiple s3 buckets?

Well, our system includes several functions, some of which operate with different buckets for the purpose of organization, permissions, etc.

For me, the use of multiple buckets was nothing special. There are several reasons why a server needs to support multiple buckets or even multiple AWS accounts.

So I think it would at least make sense if we could use multiple Uppy instances in the long run, even if they might not run optimized then. If my workaround hadn’t worked, my only option would have been to set up a second server.