https://typesense.org logo
Join Slack
Powered by
Hi guys we are having high load who can help me to...
# community-help
s
Hi guys we are having high load who can help me to upgrade capacity?
f
Are you using Typesense cloud?
s
Yes
?
f
You can hit the 
Modify
button in your cluster's "Cluster Configuration" tab in the sidebar. From there you can upgrade / downgrade your cluster's config
s
we had hight cpu can you suggest update? I will send to you cluster identifier in direct message we have high load yesterday
f
Hey there, it looks like you initiated an upgrade. It looks like the CPU spiked at 100% for a consistent period, from under 10% to 100% in the span of an hour. Did you go live at that time or is there something else that might have caused this traffic spike?
s
Now my goal is to upgrade it until I will solve issues causing high traffic. I have started upgrading from v4Cpu to v8 Should I also bump cluster version from v27.0.rc30 -> v28......
Also Ia m not wanting to have downtime as it is production
f
Yes, the 27.0rc build is months old, try upgrading to at least v28 GA or a new v29 RC build
There shouldn't be any downtime when upgrading, since you're using HA
s
Will that upgrade break something I am using node js lib "typesense": "^1.7.2" which I am guessing aslo is old
f
The caret in the in semantic versioning means that you're accepting minor / patch versions (meaning at most 1.8-1.9). We're at 2.1.0-4 currently, so try updating to that, by:
Copy code
npm i typesense@2.0.1-4 #npm
s
Ok thank you
@Fanis Tharropoulos Where I can find change log are there any breaking changes?
f
For the JS client lib or Typesense core?
s
For Typesense core
f
They're available on the 
/api
route of every major release: https://typesense.org/docs/28.0/api/
Also, there are no breaking changes between 27.0 and 28.0
s
Also I am creating embeddings for documents can it be reason why it is handling batches slowly will this cpu upgrade help or I still need gpu?
f
CPU will help, but offloading the work to OpenAI or using GPU will speed things up. Generating embeddings in general is a heavy task though, it will always be slower than not generating them
Open in Slack
PreviousNext
Powered by