sn

It's not as valuable in and of itself as monitoring SMART counters. We've only had a single SSD report failures during a long test, and it also reported an uncorrectable error. However, not finishing the test is a good proxy for if a drive is overloaded and less able to perform routine housekeeping.

Things I've learned from using SSDs at prgmr:

Since the firmware is more complicated than hard drives, they are way more likely to brick themselves completely instead of a graceful degradation. Manufacturers can also have nasty firmware bugs like https://www.techpowerup.com/261560/hp-enterprise-ssd-firmwar... . I'd recommend using a mix of SSDs at different lifetimes, and/or different manufacturers, in a RAID configuration.

How different manufacturers deal with running SMART tests under load drastically varies. Samsung tests always take the same amount of time. The length of Intel tests vary depending on load. Micron SMART tests get stuck if they are under constant load. Seagate SMART tests appear to report being at 90% done or complete, but the tests do actually run.

Different SSDs also are more or less tolerant to power changes. Micron SSDs are prone to resetting when a hard disk is inserted in the same backplane power domain, and we have to isolate them accordingly.

Manual overprovisioning is helpful when you aren't able to use TRIM.

What a drive does with secure-erase-enhanced can be different too. Some drives only change the encryption key and then return garbage on read. Some additionally wipe the block mappings so that reads return 0.

There used to be a Micro Center in the Mercado shopping center. There's no Micro Center in Northern California any more.

SF Bay Area folks: for buying new your options (depending on what you are trying to buy) are now https://www.centralcomputers.com/ for enterprise/personal computers, https://www.bestbuy.com/ for consumer electronics, https://anchor-electronics.com/ for electronic components.

Central computers just opened a store almost right across the road from the Sunnyvale Fry's.

This inspired me to minimally clean up our March 2019 SCALE17x presentation "Accidentally Accessible: a Mostly-FOSS Workflow:" https://youtu.be/FvrQPdt1X30 Unfortunately the default view is clipped very badly and doesn't include Chris' screen while he's comparing writing python in emacs versus nano, but otherwise I think it turned out all right.

We're hoping to do an updated version at LinuxFest Northwest in April, assuming the presentation is accepted.

Yes, he was. I, along with Alan Post, took over prgmr from Luke. We're hiring another developer full-time in January.

Thank you for bringing this up. The specific term, also a keyword in your linked paper, which describes this is "stereotype threat" https://en.wikipedia.org/wiki/Stereotype_threat . I'm calling it out specifically only because it's handy to have a short phrase to describe the phenomenon, and I'm not sure if everyone is aware of it or not.

It won't run out of stack space with a high enough optimization level so that it uses tail recursion. https://en.wikipedia.org/wiki/Tail_call

You can try this yourself by compiling:

#include <stdio.h>

int i = 0; int x() {--i && x(printf("x"));} int main() {x();}

using gcc with and without '-02'.

prgmr.com has an updated netboot installer: https://prgmr.com/blog/2019/10/17/openbsd.html

We rebuilt our netboot installer but we had to disable all but one of the consoles due to a race condition: https://github.com/prgmrcom/prgmr-image-source/commit/aaa8cc... This is a bug conceptually similar to https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=932416 Does anyone know offhand which nixos component we should file a bug against? Is it https://github.com/NixOS/nixpkgs ?

Sarah Newman from prgmr.com here. It's funny you should ask this - we just kicked off an effort a week ago to work on this problem. We don't even have a website yet for the software, but please feel free to watch audiplot.org for future developments.

We still do this, though the implementation has had a complete rewrite. It's very accessible.

I work with a blind developer. We create personal feature branches, file a ticket for merge requests in our ticketing system, code review them directly in the ticket or via email, and then rebase the feature branch onto master. We don't, but you could plausibly do reviews directly in git via editing the file with any comments.

I've been wanting to try gerrit however: https://gerrit-review.googlesource.com/Documentation/dev-des...

Please report back on whatever you end up doing.

One caution I have is that long term you should consider moving away from the gitlab tools if they are unable to fix accessibility concerns, so that everyone is on equal ground.

I also don't know how flexible he is on which browser or operating system he uses; he may want to try some others to see if they work any better.

I would be more than happy to help find work for anyone who is unemployed due to being "overqualified."

srn from prgmr.com here. Our tagline originally came from us being a low cost service, but I like to think of it as a customer support philosophy. One meaning is we want you to be able to fix the problem yourself by giving you instructions instead of logging into your system. Another is we try give you the benefit of the doubt that it could be our problem and not assume it's yours when there's an issue.

Checklists and written procedures are very important. One of the earlier things I did when coming into my company was create a written procedure for software upgrades until we had time to automate it with ansible.

One thing I have not had very good discipline about is I want to use checklists both for code submitted for review and when I'm doing reviews. Lint checkers etc. can only go so far.

If anyone has published checklists for code reviews I'd be curious to see them. This one seems reasonable: https://www.liberty.edu/media/1414/%5B6401%5Dcode_review_che... though I'd add concurrency to the list.

Are you asking for a recommendation on how to do this with ESXi or a recommendation for an alternative to ESXi?

The recommendation to only wipe the first and last part of the drive before reusing them is inadequate. You're best off wiping the entire drive with either zeros or hdparm security-erase-enhanced. The reason is that not all metadata is kept just at the beginning or end of the drive, especially if you have multiple partitions in the drive.

If that is untenable, mdadm has an option --zero-superblock, lvm has pvremove, and wipefs will remove file system signatures, which you can run before changing the storage layout.

I just got permission to advertise smaller subnets than a /24 during a move.

They have install times orders of magnitude less than competitors - we're talking 2 days lead time vs. 45.

And while they're not as cheap as Cogent, they are still lower cost than most other transit providers and actually have full IPv6 tables (unlike Cogent.)

Maybe that's location dependent? They'll sell us a lower commit than the full circuit, but we're in the SF bay area.

> Fortunately he was headed north on US 101, not south.

The article states: "The vehicle was traveling southbound on Highway 101 in Palo Alto."

Don't risk your family savings.

Feel out if you might be able to go part time at your current company. Half days or a day off a week may be more appealing to them than you leaving. If not the current company, then exercise your network - put out there you are looking for something part time.

Even if you are a generalist, you can look like an expert at something. I wear many hats personally.

Please do not recommend squirrelmail. It is not really being maintained.

https://www.openwall.com/lists/oss-security/2018/07/26/2

https://www.openwall.com/lists/oss-security/2018/03/17/2

https://squirrelmail.org/index.php

https://blog.cpanel.com/the-death-of-squirrelmail/

I'm not sure how programming FPGAs compares to GPU programming, but it's a completely different way of thinking compared to traditional software.

The last time I really did FPGAs was over 10 years ago, but unless the tools have gotten orders of magnitudes better, in addition to the other concepts mentioned you need to understand clock domains, metastability, pipelining, etc.

I preferred VHDL to Verilog because you shouldn't need a lint checker for your HDL. But unless it's changed, Verilog is a lot more popular.

Training is available, for example https://www.xilinx.com/training/atp.html . It probably costs more than you're interested in paying, and you should have the basics covered first.

At Lockheed Martin I worked on a SIMD processor designed before VHDL and Verilog existed. The reference design was written in a lisp-like language. I don't know if the language had a name or if it was ever publicized in any way.

I'd be happy to explore with your friend if I know of any consulting/contract opportunities for her and provide some guidance on resources for going forwards with that. She may get better traction that way; contractors are treated much differently from employees.

Why is software virtualization not an option?

Unfortunately asking for directions is not an option for everyone, to put it lightly.

https://www.npr.org/sections/thetwo-way/2018/04/15/602598119...

I believe what you're seeing is market competition, not reduction in hardware costs.

The total cost to offer compute has actually gone up significantly over the last year or so due to memory prices shooting up. I believe the cost of processors has gone up too, but not as drastically.

This seems like more of an argument against bad CoCs and against CoCs for all organizations, rather than a CoC always being bad for every organization.

There's also a difference between specifying a CoC and enforcing a CoC. Maybe there needs to be a different name for expectations that aren't going to be enforced, either because they're too imprecise or the people running things are incapable of doing so.