I was surprised to find that ~65k `swf_assets` records had their
`manifest_cached_at` timestamp set to `0000-00-00 00:00:00`? I don't
literally see where in the code that could happen. Maybe it's old?
Anyway, it's causing a problem with the utf8 database migration I'm
working on, so I manually set all of those to `NULL` and saving this
so I can test the new migration against it!
We ran the new `rails swf_assets:remove_duplicates` task in the
`impress` repository, saving our work before (previous commit) and
after (this commit), in case we need to study the results or roll back!
We're in the process of migrating away from translating these records,
because Neopets hasn't supported non-English languages in many years,
and it'll simplify our code and database lookups.
In Main DTI, we already wrote code to copy these fields onto the main
records and keep them in sync for now; now, once DTI 2020 isn't
referencing them anymore, it should be safe for the main app to drop
the tables altogether.
Note that some Prettier changes got mixed in here and that's fine!
I also wasn't suuuper careful testing these, most of them seem to be
trivially testable by just loading the homepage or doing a few basic
wardrobe actions, and the others are in Discord support log actions
that aren't enabled in development mode, so I'm just like… ehh I'll do
a couple support actions after deploy and see that they don't crash!
Note also the change to the public-data-constants, because we've
started (but not finished) moving name fields into the rows themselves
and deprecating the translations.
Weird! Well! This caused two issues.
One is that we used to filter down to only assets whose urls end in
`.swf`, because I never added support for the sound ones :p
The other is that we were using the SWF URL to infer the potential
manifest URLs, which we can't do when the SWF URL is a weird
placeholder!
Thankfully, just yesterday (wow!) we happened to have Classic DTI keep
track of `manifest_url` during the modeling process, and we backfilled
everything. So the most recent `temp` assets have their manifest! But
slightly older ones (not that old tho!!) didn't, so I manually inferred
the manifest URL from the asset's `remote_id` field instead (which
worked cuz there was no hash component to the manifest URL, unlike some
manifests).
The item "Flowing Black Cloak" (86668) on the Zafara is the one we were
looking at and testing with!
One problem I run into with the archive task is that sometimes the queries time out? My hunch is that maybe some of the assets have like, weirdly big manifest files that are being transferred as surprisingly big text files?
Anyway, I'm increasing the timeout to 20s, which is big, but big feels good for a script that doesn't run often and where failing is not great news!
I'm also idly considering whether I wanna finally put in the work to do a bulk S3 uploader sometime, because the current version that iterates over multiple `aws s3 cp` calls is just real slow, I think because it establishes a new connection each time, and that operation is maybe surprisingly expensive? And the CLI doesn't have a way to do multiple uploads any more precisely than "sync up this whole folder", which is slow when the folder contains a lot of stuff you _know_ you don't want to head up there.
Now, someone with production DB access can run `yarn db:export:public-data` to create `public-data-constants.sql` and `public-data-from-modeling.sql`.
Then, someone setting up their dev database can run `yarn db:setup-dev:full` and get all the wearables data imported right into their dev database!
I'm noticing just how poorly I'm keeping up with my own goals for finishing up DTI, and wondering if now is a good time to circle back to some old offers for code contributions I got last year… I also just figure that making this app Possible To Run with a backup of the basic public database is like. a pretty handy thing to have for archival's sake imo
Note that, for this change, we also set up Git LFS (Large File Storage). Github should be automatically compatible with this! It's a way to not write the whole 30MB database dump into the repository history, and instead keep it in a secondary filestore, because Git's core algorithms aren't really built to handle large blobs of data very well. Users setting up their dev environment will therefore also need to have Git LFS installed for this script to work! (Otherwise, they'll see a "pointer" file in `public-data-from-modeling.sql.gz` that contains some metadata about the file state but not the data itself.)
Ok great, we can now run the delta archive process!
It'd be nice to get this running on cron on the impress-2020 server, to a temporary folder? I *do* want to be remembering to run something regularly on my personal machine too though, to keep my own copy up-to-date…
That's what I get for not fully testing lmao! But right, paths in shell scripts are relative to the working directory, and if I want to be relative to the script I gotta use dirname!
Sat down and thought about the structure here and how to make the full/delta stuff make more sense together! Here's what I came up with!
In both full and delta archiving, we prepare the manifest, we create the local archive, then we upload it to remote.
I like running the full `archive:create` to help us be _confident_ we've got the whole darn thing, but it takes multiple days to run on my machine and its slow HDD, which… I'm willing to do _sometimes_, but not frequently.
But if we had a version of the script that ran faster, and only on URLs we still _need_, we could run that more regularly and keep our live archive relatively up-to-date. This would enable us to build reliable fallback infra for when images.neopets.com isn't responding (like today lol)!
Anyway, I stopped early in this process because images.neopets.com is bad today, which means I can't really run updates today, lol :p but the delta-ing stuff seems to work, and takes closer to 30min to get the full state from the live archive, which is, y'know, still slow, but will make for a MUCH faster process than multiple days, lol
Print out the image hash for easier debugging (can look up the custom data ourselves to check it), and also fix a bug with retries not carrying `contextString` through, oops!
It seems to be working!! How exciting!! I'm just letting it run on stuff now :3
One important issue is that Classic DTI doesn't show images for items modeled this way, because we don't download the SWFs for it. But I wanna update it to stop using AWS anyway and do the same stuff 2020 does, I think we can do that pretty sneakily!
I'm looking into what it would take to update the archive on a regular basis. The commands right now *are* pretty good at avoiding duplicate work… but the S3 upload still seems like it's taking very long even to just validate what's in the archive already. We might have to build our own little cache rather than using `aws s3 sync`, if we want faster incremental updates?
Here, I make a few quality-of-life changes to add a `archive:create` command that runs everything in a straight line. That way, I can let it run and see how much wall-time it takes, to be able to decide whether speeding it up feels necessary. (vs whether it's a few-hours task I can just set a reminder to manually run every week or something)
I'm working on account creation, and got tripped up by transactions not working correctly, because MyISAM tables simply don't support them! Took too long to debug this lol :p
Anyway, I made the change in the prod db, and then re-downloaded the schema files to here.
Specifically, we rename the dev database to `openneo_impress` for consistency with the main app, and create a second schema file for `openneo_id`, so we can do local account creation.
Hey this is an exciting development! A list of URLs, that we want to clone onto our hard drive, turns out to be something `wget` is already very good at!
Originally I used `wget`'s `--input-file` option to process the `urls-cache.txt` file, but then I learned how to parallelize it from this StackOverflow answer: https://stackoverflow.com/a/11850469/107415. (Following the guidance in the comments, I removed `-n 1`, to avoid the overhead of extra processes and allow `wget` instances to keep using shared connections over time. Idk why it was in there, maybe the author didn't know `wget` accepts multiple args?)
Anyway yeah, it's working great, except for the weird images.neopets.com downtime! 😅 Specifically I'm noticing that all the item thumbnail images came back really fast, but the customization images are taking for-EV-er. I wonder if that's just caching properties, or if there's a different backing server for it and it's responding much more slowly? Who's to say!
In any case, I'm keeping the timeout in this script pretty low (10 seconds), and just letting failures fail. We can try re-running it again sometime when the downtime is resolved or the cache is warmed up.
Especially in our item thumbnails, there's a lot of messiness about what the URL protocol is. There are also some SWF assets whose "URLs" are just saved as paths.
In this change, we start processing all our outputted URLs through a `sanitizeUrl` function, which tries to massage it into an `https://images.neopets.com` URL, and warns if it cannot.
This also warns on some intentionally-different URLs, like our April Fools prank item lol
Anyway, I love functions like this, because the warnings always help me discover the data problems! I wasn't aware of the path-only SWF URLs, for example, until this script started warning about the URL parse errors!
Here, we read URLs out from the swf_assets table, including SWFs, manfests, and everything referenced by the manifests.
There are a few data-polishing tricks we needed to do to get this to work! Most notably, newer manfests reference themselves, but older ones don't; so we try to infer the manifest URL from the other URLs. (Our database caches the manifest content, but not the manifest URL it came from.)
Just working on making an images.neopets.com mirror, just in case! To start, I'm extracting all the URLs we need to back up; and then I'll make a separate script whose job is to mirror all of the URLs in the list.
Hey finally! I got in the mood and did it, after… a year? idk lol
The button should only appear for outfits that are already saved, that are owned by you. And the server enforces it!
I also added a new util function to give actually useful error messages when the GraphQL server throws an error. Might be wise to use this in more places where we're currently just using `error.message`!
Right, ok, `db.close()` needs to be `db.end()` now.
This probably didn't break the user-syncing cron job though, because that doesn't automatically update I think? so it should still be comfortably running older version of the code that should still work just fine
Well, instrumentation seems to be working fine again! The bug we ran into during commit e5081dab7e is gone. Cool!
I want to be able to see what's making the new box slow. My hypothesis was (and it seems to be right) that communication with the database on the Classic DTI server is slow.
But now that they're on the same Linode account and region, I think I can set up a private VLAN to make them muuuch faster. We'll try it out!
Oh right, we have another prompt we need to not prompt for, lol :p
Here, I just assume that there's one database connection on the Auth0 account. If that's not true, the script will error—not because this is a fundamentally unresolvable problem, but because I don't want to write code for configuring a situation that doesn't exist yet :p
Huh, oops, there are a _few_ reasons the user sync cron job hasn't been running correctly.
I fixed some of the config in prod, but then discovered one more issue: the script prompts for an admin database password, so of _course_ it can't auto-run, lol.
Instead, I've now created a `impress2020-util` account, with just a few permissions (but specifically the `openneo_id.users` permission that I _don't_ give the app!), and added the username and password to the secret .env file, both locally and in prod. (In this case, prod means the Linode VPS, not Vercel, because that's where our cron runs.)
Hm, okay, so the documented way to not instrument anything doesn't actually stop them from patching Module._load. But this undocumented option sure does! So, woo, let's try it! lol
Huh, well, I can't figure out what in our production env stopped working with Honeycomb's automatic instrumentation… so, oh well! Let's try disabling it for now and see if it works.
This means our Honeycomb logs will no longer include _super helpful_ visualizations of how HTTP requests and MySQL queries create a request dependency waterfall… but I haven't opened Honeycomb in a while, and this bug is blocking all of prod, so if this fixes the site then I'm okay with that as a stopgap!
Btw the error message was:
```
Unhandled rejection: TypeError: Cannot read property 'id' of undefined at exports.instrumentLoad (/var/task/node_modules/honeycomb-beeline/lib/instrumentation.js:80:14) at Function._load (/var/task/node_modules/honeycomb-beeline/lib/instrumentation.js:164:16) at ModuleWrap.<anonymous> (internal/modules/esm/translators.js:199:29) at ModuleJob.run (internal/modules/esm/module_job.js:169:25) at Loader.import (internal/modules/esm/loader.js:177:24)
```
Oh also, this is the first time eslint has looked at scripts/build-cached-data.js I guess, so I fixed some lint errors in there.
Hope it actually work-works lol
Did some refactors in useOutfitState to support the new reset action we do after auto-saving, in case the server tweaked things like the name.
I wanted the ability to clear out closet list text for Support users, and figured I should just build the UI for end users too, and grant Support users the same access!
I narrowed down the problem to the fact that we were joining in pet types against assets, and *then* running GROUP and DISTINCT and everything. Assets x compatible species/color pairs is a LOT of rows!
Here, we instead get all the relevant body IDs first, and *then* match them against pet types—which we fetch in one batch to match body to canonical species/color.
I'm also trashing the weird caching mechanism we did here, because in practice it doesn't seem reliable anyway. If anything, I'd want to look at stronger CDN caching. (I made a small improvement to the caching annotation, but ultimately it still doesn't matter, because this query uses logged-in stuff and always comes out max-age=0 anyway.)
I already had a script for this lying around, and adapted it a tiny bit to the repository!
Part of me thought about building it in as a support tool. I might've if:
- this CLI didn't already exist
- we already had tighter permissioning, this is pretty high stakes!!
I'm not getting cron success _or_ cron failure emails for running this script on our Linode box. I was getting failures back when I had the command wrong, though.
My hypothesis is that the script output is too long to email, because of some limit somewhere along the way. I'll update the cron job to use `--skip-unchanged`, in hopes that it helps me get the emails! (I'm not suuure it's running, is the thing... though hey, here's a way to check: as of now, 512,624 of 521,896 assets are converted. If that changes eventually without a manual script run, then the cron is working!)
I want to start running this on a regular cron, and making the script faster (stop sending redundant queries) and clearer (# actually updated) is super useful for that!
