Could not load content - GET_HELPER errors

Developer help Self-hosting
1

I’m testing Ghost for a large content site with 56,000 posts and will eventually grow to 100,000. All posts are text only and not using any images. I have tried different themes but the ‘featured_posts’ always break every theme. When I disable featured_posts, everything works great, performance is snappy

Error Code: ABORTED_GET_HELPER

Error Code: SLOW_GET_HELPER

{{#get}} took longer than 5000ms and was aborted

Server Specs:

AMD Threadripper 32 Core with 128GB RAM, Gen 4 PCIE NVME
Windows 11 - Docker WSL 2
Cloudflare with local NPM proxy
Ghost CMS - 5.119.2
MySQL - 8.0.42 (database is used for Ghost CMS only)

Is Ghost CMS not designed to handle large content sites? Is there anything I can do at the database or config level to improve performance? Based on server specs, it should not take 5 seconds to fetch a featured post.

image
image1095×692 20.5 KB

@Dmitry_Spodarets @jannis @Cathy_Sarisky - Any suggestions?

2

Pretty sure the issue is not getting one featured post, but finding the featured posts in the ocean of posts.

If you run your MySQL database inside WSL (or worse: on Docker inside WSL), I’d bet WSL is the issue.

I have a 64GB RAM, Intel Core Ultra 9 185H with an NVMe drive here, and could not get decent performance with a database setup there (it was non-Ghost work, but ran a MySQL nonetheless).

WSL is basically a virtualisation. Docker is another virtualisation layer. So, either way you twist it, you will not get near-native performance, if you run this setup.

I’d try running the same workload on a native Ubuntu/Debian if you can. I’d bet that the performance is significantly better and you might get your result in under 5 seconds (for Ghost not to abort).

1 Like
3

Thank you for your response. I made a rather silly mistake with MySQL config - The default “innodb-buffer-pool-size” with default docker image was 128MB :frowning: I changed it to 10GB and now have 99% buffer free and the site has excellent performance!

It maybe an overkill but I added the following to my docker-compose file -

  --innodb-buffer-pool-size=10G
  --innodb-buffer-pool-instances=8
  --innodb-log-file-size=2G
  --innodb-log-buffer-size=256M
  --max-connections=500
  --tmp-table-size=512M
  --max-heap-table-size=512M
  --innodb-io-capacity=4000
  --innodb-io-capacity-max=10000
  --innodb-flush-log-at-trx-commit=2
  --sort-buffer-size=16M
  --join-buffer-size=16M
  --read-buffer-size=8M
1 Like