Why doesn’t any testing get done before updates are pushed out? I’m not talking about extensive testing, I’m talking about like just making sure it doesn’t cause major problems?
It probably was tested. I’ll enjoy reading the post mortem they’ll inevitably do. I’m going with, they tested it, but didn’t text xyz that demonstrated the bug.
One comment said that they found files on affected machines which consisted or mostly garbage and/or zeroes. This mostly happens when one of the machines pushing the update to the distribution servers has faulty memory. You can workaround those errors if you use checksums and double check everything, but for most departments that is too much work or they regard it as unnecessary.
This is a far more plausible explanation than crowstrike not testing their updates.
Modern computers are very consistent. Nearly all copy and download operations work correctly. If you verify with checksums, you will be running a test that always passes, so it feels unnecessary, so people want to skip it.
Helios Airways Flight 522 was a fatal caused by the pilots skipping a test that always passes. Three times their checklists told them to verify the pressurization switch is set to auto, which it always is during normal operations, except not this time because of a maintenance operation.
When this sort of error can happen even in the highly ordered checklists of a cockpit, its plausible to think it could happen in computing as well.
Why doesn’t any testing get done before updates are pushed out? I’m not talking about extensive testing, I’m talking about like just making sure it doesn’t cause major problems?
It probably was tested. I’ll enjoy reading the post mortem they’ll inevitably do. I’m going with, they tested it, but didn’t text xyz that demonstrated the bug.
The programmer’s fallacy: it worked on my machine!
But did it even in this case?
Hard to tell if Windows is BSODing because of your changes or because it’s just Thursday
One comment said that they found files on affected machines which consisted or mostly garbage and/or zeroes. This mostly happens when one of the machines pushing the update to the distribution servers has faulty memory. You can workaround those errors if you use checksums and double check everything, but for most departments that is too much work or they regard it as unnecessary.
This is a far more plausible explanation than crowstrike not testing their updates.
Modern computers are very consistent. Nearly all copy and download operations work correctly. If you verify with checksums, you will be running a test that always passes, so it feels unnecessary, so people want to skip it.
Helios Airways Flight 522 was a fatal caused by the pilots skipping a test that always passes. Three times their checklists told them to verify the pressurization switch is set to auto, which it always is during normal operations, except not this time because of a maintenance operation.
When this sort of error can happen even in the highly ordered checklists of a cockpit, its plausible to think it could happen in computing as well.