My FeedDiscussionsHeadless CMS
New
Sign in
Log inSign up
Learn more about Hashnode Headless CMSHashnode Headless CMS
Collaborate seamlessly with Hashnode Headless CMS for Enterprise.
Upgrade ✨Learn more
Jon Dixon

4 comments

Richard Soule
Richard Soule
Jan 4, 2024

Excellent post Jon.

I'd like to see the logs go to /var/log/ords instead of /etc/ords as that's where logs go on a Unix/Linux system and (quite often), the /var/log partition is separate so that if it fills up only logging stops and not the whole server.

2
·
·1 reply
Jon Dixon
Jon Dixon
Author
·Jan 4, 2024

Thanks for this Rich. Good suggestion, I will update the post to reflect this.

2
·
Connor McDonald
Connor McDonald
Jan 10, 2024

Fantastic read.

In terms of connection pool sizing, I highly recommend the work of Toon Koppelaars. (reference: neooug.org/gloc/Presentations/2019/Koopelea..)

which recommends pool sizing as:

min = max = initial.

The TL;DR rationale from the presentation for that recommendation is:

  • unused pool sessions are idle, they do nothing and consume minimal resources because the last thing pool-based apps do is free their resources when they give back a connection. If they dont, then you've got an app bug not a connection pool config issue.

  • the max size setting (by definition) is what you are prepared to let your server handle without undue stress, so why would you pick a min/initial less than that, because to do so means yours apps incurring the cost of creating a connection pseudo-randomly. That makes response times variable.

1
·
·1 reply
Jon Dixon
Jon Dixon
Author
·Jan 10, 2024

Hi Conor, Thank you for this insightful comment and the reference to the PPT. I have incorporated this into the post.

·