Mathew Heard

After tweaking the nginx S3 Cache (1.1Gbit/s now pre-warmed) to use `aio threads` we are now hitting 240% CPU on the queriers (many nodes nearly saturated). To anyone else running...

@liguozhong Am I correct in saying this also applies to chunk filtering in ingester (it looks like it does)? If so there should be a similar but seperate configuration for...

@cyriltovena What are you after here? Most queries in loki are slow, unless they span huge period of time (and hence can benifit from `split_queries_by_interval`, and even then theres a...

@ligozhong looks good. I look forward to a complete release. Thank you for your efforts.

I can second pr #5455 as very helpful for cases like this. It's not perfect. But works in a pinch. Not using it at the 100tb/day (aiming for 500gb/day) level...

Can we also do `hostNetwork: true` on the runner pods while we are at it? That should allow would allow for easier dind setups without messing with MTU. At-least for...

@mumoshu While I undestand that it can be done that way we are running individual runners with specific permissions scopes and resources. While single runnersets are possible, honestly I havent...

This is a blocker for usage with LKE (linode) too.

We see this same problem with the Samsung Smart Things devices. They are HA certified.

We can contribute feedback on Heiman sensors/lights (working), chuangkesafe (latch not working, motion low quality), Smart Things (immediate disconnect on join), TI CC253X based end devices (working), Digi XBee hardware...