• 0 Posts
  • 19 Comments
Joined 2 years ago
cake
Cake day: August 12th, 2023

help-circle



  • Interesting what he wrote about LLMs’ inability to “zoom out” and see the whole picture. I use Gemini and ChatGPT sometimes to help debug admin / DevOps problems. It’s a great help for extra input, a bit like rubberducking on steroids.

    Examples how it went:

    Problem: Apache-cluster and connected KeyCloak-Cluster, odd problems with loginflow. Reducing KeyCloak to 1 node solves it, so it says that we need to debug node communication and how to set the debug log settings. A lot of analysis together. But after a while, it’s pretty obvious that the Apache-cluster doesn’t use the sticky session correctly and forwards requests to the wrong KeyCloak node in the middle of the login flow. LLM does not see that, wanted to continue to dig deeper and deeper into supposedly “odd” details of the communication between KeyCloak nodes, althought the combined logs of all nodes show that the error was in load balancing.

    Problem: Apache from a different cluster often returns 413 (payload too large). Indeed it happens with pretty large requests, the limit where it happens is a big over 8kB without the body. But the incoming request is valid. So I ask both Gemini and ChatGPT for a complete list of things that cause Apache to do that. It does a decent job at that. And one of it is close: It says to check for mod_proxy_ajp use, since that observed limit could be caused by trying to make an AJP package to communicate with backchannel servers. It was not the cause; the actual mod was mod_jk, which also uses AJP. It helped me focus on watching out for anything using AJP when reviewing the whole config manually, so I found it, and the “rubberducking” helped indirectly. But the LLM said we must forget about AJP and focus on other possible causes - a dead end. When I told it the solution, it was like: Of course mod_jk. (413 sounds like the request TO the apache is wrong, but actually, it tries internally to create an invalid AJP package over 8kB, and when it fails blames the incoming request.)






  • One insanity in the following years was how they thought people still wanted their next generation diesel.

    I’ve been working for them in the 2010s with the department to organise the staff car fleet. We ordered many electric vehicles years ahead from production and planned it all around electric vehicles: Charging stations, operating distance, some hybrids for long distance, software to calculate trips etc.

    Then a few months before we needed them, they said: We overproduced on the latest diesel generation and can’t keep up with the demand for electric vehicles, so we have to sell the ones you ordered. You can either go with a Tesla (for official Volkswagen business trips!) or have the diesel for free.

    It felt like there was a hysteria: Decision makers got it in their heads that the “hype” for electric vehicles was ideology-driven and not something people with buying power actually wanted today or in the near future. Bit like the republican administration thinking that “woke” is our main problem. Meanwhile, huge research and development departments did come up with the electric vehicles they sell today (and fully working hydrogen prototypes you won’t see in a store, just to be safe) and must have been quite frustrated that so few were produced.






  • In the 90s during the first “mild hype”, I had Suse for quite a while, twice. Same problem with unavailable software though, I remember PGP Disc not being available back then. I remember the cool kids talking about Red Hat and Debian, you must have been one of them.

    Probably going back now, since my 2011 hardware won’t work with Windows 11.


    • One important metric: m³ / (hour*$), so how much it can filter for how much money
    • Also the volume on the setting that gives the filtration power needed. Often it is best to get a bigger one, run on middle or low level, especially for office or bedroom.
    • How much m³ / hour overall? When it is against dust, allergens, pm2.5 etc., filtering your room volume once per hour is decent. To protect from viral infections, e. g. at a dentist or doctor, 6x the volume of the room is ideal. For private use, it’s nice when it can do once per hour on a lower setting, and for occasional parties 6x of that on a high, loud setting to avoid spread of viruses.

    Pretty good for your money is the Corsi-Rosenthal Box mentioned already. As for things that don’t require assembly, Trotec beats all prices in Germany, e. g. the 250E or 350E: https://de.trotec.com/shop/design-luftreiniger-airgoclean-250-e.html That would provide more than you need for typical home use already. For a single room such as your bedroom, you can get something really decent for less than $250.

    The ones that are below HEPA standard are not as bad as somebody mentioned, imo. Against many things, such as dust or allergens, they should be fine. I’m buying only HEPA filters myself, though; doesn’t really save much otherwise.





  • A few countries still have somewhat precise numbers. UK has the ZOE health study, which found over 1 million people currently being infected out of roughly 50 million (from memory; I don’t know how many people live in UK). Germany has the SentiSurv study, indicating incidences approaching 1000 again. While the latter is only a survey in a few major cities, it allows calculation of a dark figure when put in relation to officially registered cases, which can then be applied to all regions that have the same criteria for when to test.

    Overall, not great that millions will miss a chance to get the upcoming vaccine that would provide very decent protection against the most common strains.