Banking, OpenWhisk, LAN, VPN, Learning, Google Sites

  • Dark patterns in banking. Yes, unfortunately that does happen. As example S-Bank in Finland. I've very clearly chosen the default payment account. But when you're using their mobile bank. They're opting default payment account as credit card. So they can charge lots of extra fees for each transaction. I've noticed to correct that everytime before once yesterday. They charged hefty fees for that. And probably I haven't even seen all fees yet, because there will be extra fees added (like invoicing fees, etc) after the transaction has completed and it's basic fees have been charged. So? Why they do not honor the default values I've personally configured? I'm pretty darn sure, I'm not the only one who has made this mistake. - It's good to remember that all deals, offers and default options, are like the worst ones you could have. Never accept anything as they offer it. - The easier they want to make it for you, the worse it is probably for you.
  • Had long discussions with friends about Apache OpenWhisk. And similar microservice platforms. It's important to consider how much fragmentation is preferred. Sometimes it can be really beneficial and awesome. Sometimes it'll make systems just extremely sluggish and highly unreliable. What's the right balance and when to use. That's a good question and needs to be balanced out case by case. But I've seen in past such solutions causing extreme slowness, without apparent and clear cause to developers. It's the very classic IPC latency issue, which can be a problem even when code is running on a single computer. In this light, I'm just wondering why microkernels or even nanokernels haven't taken over a good while ago? Those should provide obvious (?) benefits.
  • It seems that the LAN network problems I've reported and troubleshooted earlier, aren't gone yet. Just yesterday the issues hit again, seemingly quite randomly. What remains to be done? Disconnecting all network devices, and plugging those back 'on demand'. That's bad way do deal with the issue. But it might be any device on the LAN causing these issues. Also the core switch hasn't been changed. I've got suitable old switches in stock, so maybe I'll swap the switch one morning before there are too many people around to complain about the network outage. Last and hardest to replace is the edge router. But I'll try to be very wishful it's not the source of the problems. Because replacing that would be expensive and time consuming task. If changing the floor switch will fix the issue, it doesn't mean that the switch would be faulty. I've seen so many times different ethernet failure modes where noting is wrong, but two devices just won't work well together. Adding something like small switch between the devices might fix the problems completely.
  • Great VPN post by EFF. Nothing to add, but in some cases VPN can also weaken you privacy routing traffic via single choke point. Lot of things depend on your threat model.
  • Network issues continued, so I replaced whole floor switch. It's interesting to see, if this has any effect on the problems experienced. Added extra switch and split floors network devices between these two switches connected to router. It seems that the problems went away, for at least for some users. So it might be possible that some device on the network is causing these problems. Nor maybe not. More testing is still required.
  • Julia Evans Lear at Work. Awesome post. Yes, that's what I do all the time. Study, study and study more. But I study also evenings and weekends. Because if you need to actually work at work, you can't spent too much time at work studying.
  • It seems that this new Google Sites isn't dating posts automatically anymore. I'm not sure if that's good or bad. At least it might make it hard to know if some information is really out dated or not. Anyway, fact is that most of my posts come from such a long backlog / pipeline that when I'm posting something, it's already seriously outdated. It's just to know what I were up to. Not giving out current information.
  • I did hide all the individual blog posts from the navigation bar, because it was making the page load ridiculously heavy. New Google Sites also uses JavaScript heavily, which makes heavy pages extra slow to process maxing out CPU (single thread). Even on desktop pages are slow, and on mobile, ok, let's not talk about that.
  • The new Google sites also uses strange headers, which are there to destroy performance? cache-control: no-cache, no-store, max-age=0, must-revalidate - pragma: no-cache - expires: Mon, 01 Jan 1990 00:00:00 GMT - and naturally etag is also missing. I find this kind of performance unfriendly configuration. At least images got etag, as well as sane cache-control values and expires content (which matches the max-age given by cache-control), even if that's unnecessary afaik. As well as for static files they're advertising in headers the option to use QUIC with their site. It's good to notice that the dynamic content is served by server called "ESF" as well as the static content is served by "fife".
  • One more major UX and usability "feature" if we say it nicely. Delete page doesn't do any kind of confirmation and it's in the list with other options. Thats ffff...ailure. Just deleted one page totally accidentally without any confirmation. Luckily I've got full backups, but still, that's way annoying. - Thank you for that Google! Just as you can expect, the blog post list still contains the link to the accidentally deleted content, because the lists feature isn't supported by Google Sites anymore.