1668MB cache storage in chrome

[hackhat]

Even if I delete all these items the cache doesn't go down.

I went to "\AppData\Local\Google\Chrome\User Data***\Cache" and checked the whole folder doesn't increase on refresh and is not even the size of the storage chrome show. It might be a chrome bug?

config:

module.exports = {
  siteMetadata: {
    title: '***',
    siteUrl: `https://www.***.com/`,
  },
  plugins: [
    `gatsby-transformer-sharp`,
    `gatsby-plugin-sharp`,
    {
      resolve: `gatsby-plugin-sitemap`,
      options: {
        exclude: [
          '/internal/*',
        ]
      }
    },
    `gatsby-plugin-catch-links`,
    `gatsby-plugin-emotion`,
    `gatsby-plugin-typescript`,
    'gatsby-plugin-react-helmet',
    {
      resolve: `gatsby-plugin-manifest`,
      options: {
        name: 'gatsby-starter-default',
        short_name: 'starter',
        start_url: '/',
        background_color: '#000000',
        theme_color: '#000000',
        display: 'minimal-ui',
        icon: 'src/content/logo/square.png', // This path is relative to the root of the site.
      },
    },
    'gatsby-plugin-offline',
    {
      resolve: `gatsby-source-filesystem`,
      options: {
        path: `${__dirname}/src/content/pages`,
        name: 'markdown-pages',
      },
    },
    {
      resolve: `gatsby-source-filesystem`,
      options: {
        path: `${__dirname}/src/content/images`,
        name: 'images',
      },
    },
    {
      resolve: `gatsby-transformer-remark`,
      options: {
        plugins: [
          {
            resolve: `gatsby-remark-images`,
            options: {
              // It's important to specify the maxWidth (in pixels) of
              // the content container as this plugin uses this as the
              // base for generating different widths of each image.
              maxWidth: 1200,
              linkImagesToOriginal: false,
              backgroundColor: 'transparent',
              withWebp: true,
            },
          },
        ],
      },
    },
    {
      resolve: `gatsby-plugin-google-analytics`,
      options: {
        trackingId: "****",
        // Puts tracking script in the head instead of the body
        head: false,
        // Setting this parameter is optional
        anonymize: false,
        // Setting this parameter is also optional
        respectDNT: false ,
        // Avoids sending pageview hits from custom paths
        exclude: [],
        // Any additional create only fields (optional)
        sampleRate: 100,
        siteSpeedSampleRate: 100,
        cookieDomain: "www.***.com",
      },
    },
  ],
}

Description

1668MB cache storage

Looks like every refresh adds 10MB. Doesn't look like is based on some kind of time (the more you wait the more it adds). Just on every refresh I get more 10-20MB of cache.

Steps to reproduce

I don't know

Expected result

certainly less than 1.6GB...

Actual result

1668MB cache storage

Environment

System:
OS: Windows 10
CPU: x64 Intel(R) Core(TM) i7-5600U CPU @ 2.60GHz
Binaries:
Yarn: 1.9.4 - C:\Program Files (x86)\Yarn\bin\yarn.CMD
npm: 5.5.1 - C:\Program Files\nodejs\npm.CMD
Browsers:
Edge: 42.17134.1.0
npmPackages:
gatsby: ^2.0.19 => 2.0.21
gatsby-image: ^2.0.17 => 2.0.17
gatsby-plugin-catch-links: ^2.0.4 => 2.0.4
gatsby-plugin-emotion: ^2.0.5 => 2.0.5
gatsby-plugin-google-analytics: ^2.0.6 => 2.0.6
gatsby-plugin-manifest: ^2.0.6 => 2.0.6
gatsby-plugin-offline: ^2.0.5 => 2.0.6
gatsby-plugin-react-helmet: ^3.0.0 => 3.0.0
gatsby-plugin-sharp: ^2.0.8 => 2.0.8
gatsby-plugin-sitemap: ^2.0.1 => 2.0.1
gatsby-plugin-typescript: ^2.0.0 => 2.0.0
gatsby-plugin-typography: ^2.2.0 => 2.2.0
gatsby-remark-images: ^2.0.4 => 2.0.4
gatsby-source-filesystem: ^2.0.3 => 2.0.3
gatsby-transformer-remark: ^2.1.7 => 2.1.7
gatsby-transformer-sharp: ^2.1.5 => 2.1.5

[KyleAMathews]

Can you reproduce this with the default starter?

[hackhat]

Only happens when deployed and not in localhost... Check the updates:

Even if I delete all these items the cache doesn't go down.

image

I went to "\AppData\Local\Google\Chrome\User Data***\Cache" and checked the whole folder doesn't increase on refresh and is not even the size of the storage chrome show. It might be a chrome bug?

[hackhat]

@kakadiadarpan what more info do you need?

[kakadiadarpan]

@hackhat Is it possible for you to provide us with a reproduction repo?

[krismorf]

I also notice a similar issue:

1. go to https://5bd3136f02ed8308afdcbfc5--gyftshop.netlify.com/
2. open console
3. start browsing

You'll see that each page takes around 3-4MB of application storage.

[hackhat]

@kakadiadarpan sorry I can't provide production urls.

[hackhat]

@kmorf can you please list all your gatsby plugins?

[krismorf]

System:
    OS: macOS 10.14
    CPU: x64 Intel(R) Core(TM) i5-6360U CPU @ 2.00GHz
    Shell: 5.3 - /bin/zsh
  Binaries:
    Node: 11.0.0 - /usr/local/bin/node
    npm: 6.4.1 - /usr/local/bin/npm
  Browsers:
    Chrome: 70.0.3538.77
    Safari: 12.0
  npmPackages:
    gatsby: ^2.0.31 => 2.0.31
    gatsby-image: ^2.0.17 => 2.0.17
    gatsby-plugin-canonical-urls: ^2.0.6 => 2.0.6
    gatsby-plugin-create-client-paths: ^2.0.1 => 2.0.1
    gatsby-plugin-feed: ^2.0.8 => 2.0.8
    gatsby-plugin-google-analytics: ^2.0.6 => 2.0.6
    gatsby-plugin-lunr: ^1.2.0 => 1.2.0
    gatsby-plugin-manifest: ^2.0.6 => 2.0.6
    gatsby-plugin-netlify: ^2.0.2 => 2.0.2
    gatsby-plugin-netlify-cache: ^1.0.0 => 1.0.0
    gatsby-plugin-netlify-cms: ^3.0.4 => 3.0.4
    gatsby-plugin-offline: ^2.0.9 => 2.0.9
    gatsby-plugin-react-helmet: ^3.0.0 => 3.0.0
    gatsby-plugin-robots-txt: ^1.3.0 => 1.3.0
    gatsby-plugin-sharp: ^2.0.8 => 2.0.8
    gatsby-plugin-sitemap: ^2.0.1 => 2.0.1
    gatsby-plugin-styled-components: ^3.0.0 => 3.0.0
    gatsby-remark-images: ^2.0.4 => 2.0.4
    gatsby-source-filesystem: ^2.0.5 => 2.0.5
    gatsby-transformer-remark: ^2.1.9 => 2.1.9
    gatsby-transformer-sharp: ^2.1.5 => 2.1.5
  npmGlobalPackages:
    gatsby-cli: 2.4.3

[ryanwiemer]

@davidbailey00,

Are the other offline issues that you were looking into possibly related to this problem of the cache increasing on every refresh?

It looks like you just submitted a PR to not precache the index page. (#9603)

[ryanwiemer]

@KyleAMathews,

I haven't been able to reproduce it with the default starter but it is pretty apparent on https://www.gatsbyjs.org/. Each refresh adds about ~10mb to the cache.

[hackhat]

Before cleaning:

After cleaning:

I'm inclined to think is a bug in chrome as it doesn't free space on disk.

Anyway, we should investigate other browsers; Edge doesn't show this data.

Also, maybe we can have some automated tests to make sure it won't happen in the future if is a bug with gatsby.

[vtenfys]

Are the other offline issues that you were looking into possibly related to this problem of the cache increasing on every refresh?

No, the other problems are all unrelated - the only reason to disable precaching the index page is because it's runtime cached anyway, and precaching it causes additional problems with POST requests.

I'm guessing that Chrome isn't deleting old files from the cache after each time the website is updated - maybe this is something we need to investigate. Is the list of files in the dev tools getting longer each time the cache storage increases? If so, it might be a bug on our side. If not, it sounds like a bug in Chrome, since there's no reason the cache size should increase if no additional files are cached.

[vtenfys]

I'm also unable to reproduce this on gatsbyjs.org - I've cleared all site data, then done 20 reloads in a row and the cache size doesn't change. So it looks like either 1) the bug is platform-dependent (I'm using C70 on Ubuntu), or 2) it relies on visiting some other pages first, or 3) it relies on the website being updated, i.e. with resources at different locations (so they'll need to be cached again).

[hackhat]

@davidbailey00 to me doesn't look like it adds new files to cache in the dev tools.

[vtenfys]

to me doesn't look like it adds new files to cache in the dev tools.

In that case I strongly suspect this is a bug in Chrome (possibly platform-specific) rather than a problem on our side.

Because of this, along with the fact I haven't been able to reproduce, I'll close this issue for now - if anyone has more information or another reproduction, then I'll be happy to reopen and investigate.

[hackhat]

@davidbailey00 I think we should be sure that is the case. I think is a pretty big deal if is it actually happening both in Gatsby or in chrome, so we should not rule them both out.

Because of this, along with the fact I haven't been able to reproduce, I'll close this issue for now - if anyone has more information or another reproduction, then I'll be happy to reopen and investigate.

Well, I and another user were able to reproduce, so I think is a bit too early to close it. Even if is dependent on platform, it might be a bug.

Can anybody get in touch with Chrome team to ask what they think?

I think we should keep it open.

[DSchau]

It's fair to re-open this. Both @pieh and I were able to reproduce the issue, so it seems to be a real issue.

It doesn't seem to be anything introduced super recently, as gatsbyjs.com exhibits the issue and that's on 2.0.5 of gatsby-plugin-offline. It's possible it's still a Chrome bug, but I think this deserves another look. Thanks to everyone for reporting/finding the issue!

Also note that clearing the storage does not fix the issue (at least for me). I'm able to reliably reproduce the issue even after clearing the storage. It grows by approximately ~10MB each time.

[vtenfys]

Hey there, sorry for closing this initially! It turned out that I had my adblocker enabled, which prevented this problem from occurring, which is why I couldn't reproduce the problem initially.

I've identified the cause of the problem - Google Analytics loads a tracking GIF with a unique URL on each reload, each of which is cached by the offline plugin. Since these are cached opaquely, the data is padded out which increases the storage space considerably.

To fix this we need to prevent resources like this from being cached - for resources which support CORS, including the tracking GIFs, we can get the cache-control header and detect whether no-cache is specified (note: this is unrelated to our ?no-cache=1 parameter which is soon being removed).

For resources which don't support CORS we have a bit of a problem since we can't detect whether these should be cached. As a result, I think it's best that we always use CORS when caching resources, and ignore any errors related to access control headers - this is what we were previously doing so I'll revert some recent changes.

Update: I've found a better way to fix this, by leveraging Workbox's auto caching! A fix should be published shortly :)

[alex-p-chan]

If anyone else is having the issue with large caching.
I believe this might be due to the urlPattern not being focused enough.
The service worker was caching analytics scripts and other third party scripts which it shouldn't and was blowing out my cache.
I managed to make my cache 1/100th of the size by changing the options in gatsby.config as below. obviously change mysite.com to your URL.

    {
      resolve: `gatsby-plugin-offline`,
options: {
        dontCacheBustUrlsMatching: /(\.js$|\.css$|\/static\/)/,
        runtimeCaching: [
          {
           urlPattern: /(\.js$|\.css$|\/static\/)/,
            handler: `cacheFirst`,
          },
          {
            urlPattern: /^https?:\/\/(www\.mysite\.com|localhost:8000|localhost:9000|staging\.mysite\.com).*\.(png|jpg|jpeg|webp|svg|gif|tiff|js|woff|woff2|json|css)$/,
            handler: `staleWhileRevalidate`,
          },
          {
            urlPattern: /^https?:\/\/fonts\.googleapis\.com\/css/,
            handler: `staleWhileRevalidate`,
          },
        ],
      }
}