Which models are you using for this?
28.01.2026 08:22
👍 0
🔁 0
💬 1
📌 0
Which models are you using for this?
Have you find the root cause?
Hah, might not even be a #Tahoe -only issue: old.reddit.com/r/LogitechG/...
In today's news w/r/t #Tahoe upgrade: #Logitech G Hub (and likely Logi) has stopped working. I don't really use this software, but it's a requirement to disable the terrible LEDs on my mouse and to program buttons.
3/ I'll try to go trough the workday without spending more time on this. tl;dr: Do NOT upgrade to #Tahoe yet if you have critical workloads/workflows in #Fusion VMs.
2/ On the internets, people were recommending disabling VoiceOver, but I've never even enabled that. I finally got my VMs to boot and access them via Fusion's window after finding and force quitting a leftover vm process in Activity Monitor. Not yet sure if the problem is gone for good.
1/ Upgraded to MacOS #Tahoe a couple of days ago, and had problems with the old 13.x version of #VMWareFusion. Upgraded to 25H2, which was quite terrible as you need a new binary from the Broadcom website. Then I had problems again. Machines seemingly running, but you cannot even see their windows.
Are there any plans to open source it? Or maybe blog about the schema, queries and the agent?
No defragmentation? Pass!
One of my best tech investments. Anything you’d like to know?
bsky.app/profile/kakk...
Running a Fusion VM on Apple Silicon to have best of both worlds.
Git + Nix + home-manager
Wow, is that your new PC?
Is this with Tahoe?
Any plans for Linux support?
How’s the software?
Natively?
Helm and helmfile. Anything else will be more expensive on the long run with marginal gains.
Why is it more likely that this is a migration and not for example an upgrade or other maintenance task? Or maybe you think of these as subcategories of migration?
Well, push model is also really poor in many cases, even in case that you are mentioning, it’s much easier to DDoS your backend with push than it is with pull.
Couple of comments:
- You can still get up metrics via prometheusreceiver
- Converting dots to underscores and adding suffixes is usually just a config option away
Not sure if I read the perftest results properly, but e.g. the basic counter test is 8x faster in OTel than in Prometheus?
Even though OTel instrumentation is getting over some of its problems, we are also Prometheus-first in terms of instrumentations internally. You can still bridge/OTel collect everything into an OTel backend, if needed.
How’s everyone managing their selfhosted #ClickHouse instances on Kubernetes? Bitnami/clickhouse or their clickhouse-operator? Altinity Operator? What schema management and scaling out/sharding look like operations-wise?
Do you have to opt-in for this?
How deeper discussions are working out in channels with their "thread" implementation?
It was a bit weird at first, but my HHKB's really grown on me.
It’s clear, OpenAI is running on Heroku.
If it's good ol' PHP, you don't even need FTP, just vim.
It's terrible name, yes, but why do people keep having this configured like this? Aren't they checking the logs to see errors, or - even better, aren't they concerned why the metrics are not showing up in the desired location?