Jerry 🦙💝🦙Well, backblaze just sent me a mail about their $335M deal with Coreweave, so I guess Backblaze is about to get ridiculously expensive like all the other IT services and products lately 😞
| Location | Stockholm, Sweden |
| Signal | forst.31 |
Forst
@forst
- 150 Followers
- 242 Following
- 4.2K Posts
Emperor penguin 🐧🇦🇶 ∙ Servers, networks, IPv6, Python, Linux, radio, cycling ∙ 30 ∙ He/him
Niléane
Odoben
Neil CraigA colleague pointed out that WhatHIFI measured latency on TV streaming, with a view to watching live sport.
That reminded me that we reduced the stream latency on BBC iPlayer over the last couple of weeks. You'll mostly have approx. 12 seconds latency on iPlayer TV - HD or UHD, they're the same now.
There's a few exceptions but for most people, that's where it's at and is thanks to a ton of work from BBC R&D and teams adjacent to mine in Public Service.
ROllerozxaLast night I had a dream that Microsoft pushed out a Windows update to "improve localisation" that (accidentally?) added more drive letters in latin languages with more letters that appear after Z in the collation order, but only if the system is set to that language!
So I went onto dreamy YouTube and watched a video about someone setting their system language to Swedish, then plugging in flash drives into a big tree of USB hubs until drives assigned to Å:\, Ä:\ and Ö:\ appeared in Explorer.
Rootless Podman Quadlets + systemd socket activation are such an elegant combination!
Running nginx on the host side as a reverse proxy towards unix sockets created by systemd, which get passed as open file descriptors into respective containers. No need to think about network namespace translators (like pasta or slirp4netns), bind mounts or UID/GID mapping.
… but then I also had SSH connections die on multiple occasions, which already have aggressive TCP-level and SSH-level keepalives every 5 seconds.
The timings don't look particularly even/pretty either
* 6 seconds since last ARP exchange
* 2 min 7 sec since last TCP outgoing packet
* 7 min 50 sec since last DHCP exchange
NTP queries happen every 5 minutes.
DHCP lease is 1 hour.
My main router has such traffic all the time, that's likely why I don't see that problem there.
But on a host that's mostly idling now it's much more noticeable.
Continuing the "periodic packet loss" saga. Set up a more predictable sniffer with two USB NICs and a regular Linux software bridge with all offloading disabled.
Reproduced the problem, packets occasionally stop arriving from the BNG.
Seems to be triggered by a lack of "useful" traffic from the affected client. As soon as an NTP request comes from an affected client, ICMP immediately resumes. ICMP thus doesn't count as "activity".
Same exact thing on a capture from a few days ago too.