2 ms·

> “Looks good let’s ship it and monitor logs” The best part of that approach is that it is completely unable to deal with anything that isn't logged in the fir

> “Looks good let’s ship it and monitor logs” The best part of that approach is that it is completely unable to deal with anything that isn't logged in the first place, such as bricking a device.[1] Similarly, it can't collect good info on performance issues such as intermittent lockups or freezes. NVIDIA drivers for example collect an absurd amount of telemetry, but to this day their drivers freeze my computer completely (can't even move my mouse!) for about 10 seconds a short while after logging in. No log, no problem. [1] Oh sure, you're about to say that they can log the "started patching" event and then monitor for "didn't come back up" event afterwards. They could, but they don't. For years I had issues with blue-screens after forced semi-annual upgrades, and Microsoft was blithely unaware.

eddythompson80 2y ago
Oh yeah. I absolutely agree. The answer from management (or even your direct dev manager) is “we should log more”. It absolutely doesn’t matter how much you’re logging because there is always gonna be a drill-down question you can’t answer. “log more” is almost always the answer. At my current work we’re at a place where every function call has boilerplate of func someBullshitFunction() { logger.info(“started”); defer performance.Meausure(“someBullshitFunction”) … }
- temporallobe 2y ago
  There are definitely frameworks, languages features, and libraries that reduce the need for such tedious and redundant boilerplate code.
  - eddythompson80 2y ago
    True, but how often do you get to change your company’s/team’s/product’s entire framework/language/library because you got an ask vs “ok, just bolt on this call whenever you’re doning I/O”
anytime5704 2y ago
No offense, but this is wrong. There is a metric-fuck ton of telemetry on a windows machine (really any sufficiently sophisticated software/device). Not only do they correlate the update start event with the update end event, they also have a billion other metrics that nosedive if the machine can’t actually reboot. That doesn’t make it any more acceptable, but it’s the truth. It’s totally possible the first metric report doesn’t show up for a day, too. So, it’s doubly not ok for them to push something so haphazardly.
- eddythompson80 2y ago
  How do you think they caught this in a “preview update” then?
kevingadd 2y ago
The 10 second freeze is infuriating. It's been driving me crazy and had me convinced my PC was busted.

eddythompson80 2y ago

Oh yeah. I absolutely agree. The answer from management (or even your direct dev manager) is “we should log more”. It absolutely doesn’t matter how much you’re logging because there is always gonna be a drill-down question you can’t answer. “log more” is almost always the answer. At my current work we’re at a place where every function call has boilerplate of func someBullshitFunction() { logger.info(“started”); defer performance.Meausure(“someBullshitFunction”) … }

temporallobe 2y ago
There are definitely frameworks, languages features, and libraries that reduce the need for such tedious and redundant boilerplate code.
- eddythompson80 2y ago
  True, but how often do you get to change your company’s/team’s/product’s entire framework/language/library because you got an ask vs “ok, just bolt on this call whenever you’re doning I/O”

anytime5704 2y ago

No offense, but this is wrong. There is a metric-fuck ton of telemetry on a windows machine (really any sufficiently sophisticated software/device). Not only do they correlate the update start event with the update end event, they also have a billion other metrics that nosedive if the machine can’t actually reboot. That doesn’t make it any more acceptable, but it’s the truth. It’s totally possible the first metric report doesn’t show up for a day, too. So, it’s doubly not ok for them to push something so haphazardly.

eddythompson80 2y ago
How do you think they caught this in a “preview update” then?

kevingadd 2y ago

The 10 second freeze is infuriating. It's been driving me crazy and had me convinced my PC was busted.