Another Day, Another Windows Update

You know, even though I don’t use windows a huge amount, I’d think that I’d use other programs enough to bump Windows Update off my “most recently used” list in the start menu. Nope, it’s still there. Second from the bottom, but it’s still hanging in there. I’m happy to see yet another Windows Update, this time involving among other “allow an attacker to compromise your Windows-based system and gain control over it” updates, a font.

A font update that may cause me to restart my computer.

What the hell? Do they attach that warning to every update these days, or does repacing a font seriously mean that I have to reboot.

Under linux you reboot when you change hardware, update the kernel, or something is completely fucked up and it’s causing a hang (normally driver issues, doesn’t happen that often, but it does happen).

Under windows you reboot when you change a font. <shaking head> There’s that old windows joke about “you moved the mouse, please reboot to continue”… suddenly it’s not so funny anymore.

Microsoft, you there? Paying attention? This is why people don’t like you. +1 for security updates, -1 for a font being a “critical update”, -1 for making the user reboot afterwards. That’s a sum total of -0.99995295234 according to the windows calculator. Hello? Is this thing on? I can only pray that when longhorn arrives it doesn’t suck this badly.

16 Comments on “Another Day, Another Windows Update”

  1. Wanna know what’s better? I installed the thingie, and my laptop locked up twice in the matter of about three hours. First lockup was about an hour and a half after install, second one was an hour and a half after that.
    I’ve uninstalled it to see what happens… but jeezus. Can’t they patch their patches?!

  2. Arc: about “9s” availability. The problem is that you blow your “5 9s” availability when you reboot… however, you are most likely doing it at some obscure time of day when the computer has little or no real activity. Sure, you are taking it down and “losing nines”, but what happens on the off chance that your update fucks up something and denies you the ability to fix it when it really affects production?
    If you really want “5 9s” then you need to get big iron: AS/400s, AS/390s, redundent failover servers, etc. etc. etc. If you don’t have those things, then you aren’t REALLY worried about “9s”. If you do have those things, then rebooting will not effect you because of the various failover mechanisms that you have in place.
    As to Cat5’s astute observation that in the *nix world, stopping/updating/restarting on a SERVICE level is the Way Things Are Done. Yes, this is true, but it is good to reboot every now and again anyhow.
    At a job I had at Ford Motor, All of the HP/UX machines had at most 182 days of uptimes…. not because they would crash, but because it was policy that every server be rebooted twice a year, just so that if there WERE ever a problem with booting, you would find it out sooner than later…..
    In other words, uptime is not the real goal, nor are “9s”, but rather the reduction of “unscheduled downtime”, which is the REAL bane of a business.

  3. You most likely don’t have to reboot in order for the change to instantiate, but I personally think it is a good policy. When you change something major, reboot and make sure everything works. That way, if something is screwed up, you know it immediately (when it is planned) rather than at some distant time in the future, when everything is fux0red, you can’t remember what you have changed, and it is unplanned/unscheduled.
    Every Linux person I know has been stung by the machismo of not rebooting (and no lip from you, Arc, I remember SEVERAL instances!). I appreciate that MS pretty much forces you to do it, but I think it is an enforcement of a “best practice” more than a technical requirement.

  4. Happy to see that you have not been scobleized šŸ˜‰
    I had some concern since Robert Scoble visit.
    Anyway I disagree with your “+1 for security”. Microsoft have been informed of this security issue six months ago and fixed it only yesterday (I guess when the issue was going to be disclose by the security reseach team).,1282,62239,00.html
    “Microsoft: Oops! We Did It Again”
    “Maiffret complained that the delay between eEye’s discovery and Tuesday’s public disclosure by Microsoft was “just totally unacceptable” because Windows users were broadly vulnerable during the period.”
    Microsoft: -1 for security

  5. Christophe – I can almost still give it to them, as there is testing, coding, more testing, etc that (in theory) goes on. With a company that big I would hope that they didn’t fix it yesterday, but in fact fixed it months ago and the fix only was released from testing yesterday. Am I reaching maybe? šŸ™‚

  6. maybe.
    IMHO most security fixes don’t have broad impact on the code unless it’s a design flaw. Ie. you don’t need 6 months to fix a buffer overflow.
    I applied the patch on a computer yesterday and this morning I found the windows update icon on the very same computer informing me of the availability of a new update (for Media Player IIRC). To me this indicates that at least one of the two upgrades was not tested for 6 months.
    And the fact that both patches were made available outside of the supposedly monthly schedule, seems to indicate that both patches were done in a hurry.

  7. Del – the only problem with that idea (reboot after major changes) is that it kills your availability. If you’re trying to run a mission critical server and have to reboot twice a week, or twice a month, or however often these patches come out, you’re definately not going to make your two or three or four 9’s of availability. You could just wait and reboot in a slow time, but when is that if you’re a huge multinational company? Or just wait until a few have accumulated, but then you end up with the original problem, reboot and something goes wrong, what broke it.

  8. I essentially agree with you, Del.
    My suspicion is that many of those “5 9’s” SLAs will permit the provider of services to schedule some downtime (ie, for upgrades/maintainance) for several hours if they give lots of notice.
    I logged into an HPUX box I have access to. Uptime is 297 days. That’s the core database/application server for one of my clients. I think one of the differences is that HPUX/AS400 have way less updating going on, whereas Win32/Linux systems are always getting new kernels, new patches, and so on. My impression is that people with the big iron tend to just NOT TOUCH it unless they really have to. Of course, this box doesn’t have a responsible/competent admin. No backup/testing server, questionable security, bizarre partitioning (100 mb for /home??? wtf), and so on. Not many people interested in HPUX anymore.

  9. 100 mb? Why in my day we had 50 and we liked it! šŸ™‚
    I also get the impression that the big iron isn’t generally hooked up to the net so the security isn’t so much of an issue (or at least, not the same way that it would be for say, any windows box hooked directly to the net).

  10. Arc & Del – Isn’t that one of the nice things in the *nix world though? that you can patch, then restart a SERVICE, rather than having to reboot the whole kernel/os?!?
    Microsoft is simply lying (pardon the pun) in the bed it made when it declared “everything” to be essential to the kernel… now, if you get an IE fault, it can compromise the whole system, as “IE is part of the Kernel!”
    MS gets an ‘A’ for effort, but an ‘F’ for the direction they’ve taken. If only they spent the time they now waste on “how can we lock everyone out”, on actually innovating! I think they would truly be unstoppable (and less loathed by people in the know).

  11. Why innovate when you own 90% of the desktop market?
    Sure, longhorn is just around the corner. Any day now. Honest. Totally new security model. No bugs. It’ll make your teeth whiter. Just a couple more years, honest, it’ll be great. No, don’t upgrade to anything else, it’s coming Real Soon Now(tm).

  12. While these reboots are annoying (needed?), depending how important a system is you can still keep uptime by load balancing.
    Many people forget this, but even with W2K’s built-in NLB at work we have been able to keep a good uptime, even when the Intel RAID controller required is to rebuild the whole friggen RAID to do a bios update…

  13. What’s stupid after installing that new font, is that you get asked “… would you like to reboot now to complete this… ?” with the options Yes or Cancel. What kind of stupidity is that? Should either be “Yes or No” or “OK or Cancel”, but not a bizarre questionable mix.

  14. Wim – Microsoft isn’t known for it’s sane UI or button naming conventions. I’m seeing that much more now I’m spending time in OS/X, which while it’s not perfect, it’s much better. The notable exceptions funnily enough are the elements of MS Office / X šŸ™‚