Hacker Newsnew | comments | ask | jobs | submit | ismarc's commentslogin

Our use-case is different (not cassandra or db hosted on ephemeral drives), but what we've found using AWS for about 2 years now is that when an availability zone goes out, it's either linked to or affects EBS. Our setup now is to have base-load data and PG WAL files stored/written to S3, all servers use ephemeral drives, difference in data is loaded at machine creation time, AMI that servers are loaded from is recreated every night. We always deploy to 3 AZs (2 if that's all a region has) with Route 53 latency based DNS lookups that points to an ELB that sits in front of our servers for the region (previously had 1 ELB per AZ as they used DNS lookups to determine where to route someone amongst AZs and some of our sites are the origin for a CDN, so it didn't balance appropriately...this has since been changed) that is in the public+private section of a VPC with all the rest of our infrastructure in the private section of a VPC (VPC across all 3 AZs). We use ELBs internal to the AZ for services that communicate with each other. The entire system is designed to where you can shoot a single server, a single AZ or a single region in the face and the worst you have is degraded performance (say, going to the west coast from the east coast, etc.).

Using this type of setup, we had 100% availability for our customers over a period of 2 years (up until a couple of weeks ago where the flash 12 upgrade caused a small amount of our customers to be impacted). This includes the large outage in US East 1 from the electrical storm, as well as several other EBS related outages. Overall costs are cheaper than setting up our own geo-diverse set of datacenters (or racks in datacenters) thanks to heavy use of reserved instances. We keep looking at the costs and as soon as it makes sense, we'll switch over, but will still use several of the features of AWS (peak load growth, RDS, Route 53).

The short answer is to design your entire system so that any component can randomly be shot in the face, from a single server to the eastern seaboard falling into the ocean to the United States immediately going the way of Mad Max. Design failure into the system and you get to sleep at night a lot more.

-----


You've obviously never had any real interaction with someone in the grip of a mental health emergency. She was put on suicide watch and evaluation, not because it's just fun. She cut the shit out of her leg, to where it bleed through, had repeated, documented thoughts of suicide and then tried to just pretend everything was OK. The first night was emergency care, i.e., make sure she didn't die. The next days were mental health evaluation while making sure she didn't die. You cannot be forcibly admitted in the U.S. she admitted herself. She did not have to stay after the first night, but they will make you think you do because they want you alive and to help you get better. It's not some giant conspiracy to do harm to people.

Does mental health services suck pretty much everywhere? Yes. But nothing she went through was inhumane (I've been through similar scenarios in less pleasant medical facilities) and all of it is geared to try and help.

-----

jwise0 80 days ago | link

There is plenty to moralize in your post, but I'll refrain; I'd like to factualize instead.

In many parts of the the U.S., it is possible to be forcibly admitted to a mental unit. In California, for instance, this process is called a 5150 (named for the law that allows it), and authorizes a 72-hour hold for psychiatric care. (It is also possible to authorize a 14 day hold, called a 5250.) This is not unique to California. She almost certainly did have to stay after the first night.

I want to address as well your comment as to "thoughts of suicide". The terminology for this that's used in the literature is called a "suicidal ideation", and they run the gamut from thinking about it for a moment all the way up to making plans. A suicidal ideation is (obviously) a risk factor for suicide, but is not at all a guarantee that a patient will make an attempt on their life. (To hammer the point home, there are plenty of other risk factors that we do not hospitalize for; many patients with personality disorders will try to take their own life at some point, but we obviously cannot hospitalize them all at all times, nor should we!)

Neither I, you, nor anyone outside of this person's care network will have access to her files, so it's not reasonable for me to make a judgement call on whether she should have been placed on suicide watch or not. Given only the data provided, though, it is certainly not a slam dunk that she needed to be admitted.

I appreciate the agony you must have witnessed in someone who needed urgent psychiatric care ... but the experience that you witnessed is not universal.

-----

DanBC 80 days ago | link

I've had extensive experience caring for someone with severe and enduring mental illness. That person had several stays at MH hospital, some as an informal patient and some as a patient detained under various sections of the mental health act. In about 6 years that person's DSH was severe enough to require surgical treatment and inpatient admission to general hospitals or hospitals with specialist services. (Eg burns and plastics wards). I have personal experience of accessing specialist mental health services. All of this is in the UK.

I can confidantly say that you are wrong when you say I have no experience of people in acute psycjiatric distress.

Locked rooms is very unusual. Locked wards are not normal. A person might need supervision to leave the ward but they are not prisoners and are not treated like prisoners - risk is managed by closer supervision (I know a person who had 2 members of staff within arms reach at all times) not by locking them in a room. There are rare exceptions to this with special soft rooms - all blue with heavy crash mats. In gloucestershire this room is in the low secure forensic unit, which is in the grounds of but seperate from the main adult mental health hospital for the county.

A person at severe risk of self harm or suicide will be able to do so in a very bare room - smashing their head on the walls or floor, using their clothing as a ligature.

Strip searching patients is inhumane. Intimate searches of patients is just bizarre. Especially when we remember the overlap between people who have a mental health problem and peole who have been sexually abused. Being stripped and intimately searched is distressing for most people, but could be especially so for victims of sexual abuse.

The UK charity Mind recently did a report about restraint for aggressive patients and for patients who needed rapid tranquelisation. This is something that should be used as a last resort in very clear situations. That report talked about the need to protect people from unneccessary restraint. I mention this because even this clearly protective measure (even most service users recognise a need for appropriate restraint) is looked at carefully to see if there are safer kinder alternatives.

We can't tell from her post how severe her wound was. She doesn't mention any drips, so we don't know if they needed to give her fluids or not. We don't know if they used sutures to close the wounds or if they ised steri-strips. They didn't admit her for surgery.

If her story is true it is shocking.

You say that you've experienced similar or worse. I am very sorry you went through that, and I am angry other people who should have been caring for you put you through that.

(I've noticed a couple of people mentioning my tone so I am trying to work on that. Sorry ifthis post sounds aggressive or grumpy, it isn't meant to. That's just my poor use of English).

-----

marvin 80 days ago | link

Thanks for setting the record straight on this. It makes me really sad to read all the comments here that have no clue how treatment of psychiatric emergencies should be done. If this is representative for the United States, I'm sure as hell glad I don't live there. Never been admitted to a psychiatric hospital myself, but I have enough friends who have (serious self-harm, attempted suicide by overdose or firearms). Their stories are like yours. I have never heard of anything resembling the stuff portrayed in this story. Seems to me that the US conflates violent criminals and mental patients, with some very bizarre and ethically horrendous results.

-----

Daniel_Newby 80 days ago | link

Psychiatric wards in the U.S. are for very short term stays by people who have extreme problems. They take people who are at major risk of causing harm, and patch them up just enough so that they can be thrown out. And even that is being scaled down and sped up.

The inpatient mental health system is like a mental ICU: designed for prevention of death, not for anyone's convenience.

-----

marvin 79 days ago | link

My point is that this approach causes harm. It's like treating heavy bleeding by applying a tourniquet and sending the patient out the door. It's medically unsafe, unsustainable patient care. This is not simply a question of budgets, it is a question of how society views mental health. This kind of treatment indicates an "us and them" mentality where mental patients are viewed as second-class citizens. Your wording, "extreme problems" also hints at such a dichtomy, although I'm sure you didn't do it on purpose. There are other ways to say this: Basket cases, crazy, insane, psycho, major issues, etc. A neutral term would be "very ill" or something like that.

My examples of attempted suicide by overdose or "self-influcted gunshot wounds" definitely qualify as "extreme problems", so it is clear that there is a different way to do this.

I am aware that your comment probably just meant to say that the system is set up in an unfortunate way, and that you probably don't represent the views I describe here. But from my perspective it appears to be much worse than just an underfunded system.

-----


I was on a medication that did something similar (I believe it may have been depakote or gabapentin). Thing is, I have absurdly good relative pitch, and combined with an 'internalized' sound of 60Hz (I've worked around a large number of generators and high voltage equipment over the years), I have the same net effect as perfect pitch (60Hz is roughly a quarter step above B-flat, a quarter step below B).

What bugged me was the intervals were off. It's hard to describe, but, for those not familiar with how the notes are laid out, 60Hz is quarter step below B, 120 is quarter step below B, 240 is quarter step below B, and so on. However, as you get higher, the frequency difference gets larger (B is actually 61.75ish Hz and B flat is 58.25ish Hz, next higher is 123.5ish Hz and 116.5ish Hz, and so on). Things were translated ~1/4 octave higher, which completely threw everything off because the 60Hz sound no longer fell evenly between two notes or even directly on a note, it was an 1/8th step away. This meant that as the notes got further away from 60Hz (higher or lower), the distance to the note got larger. So while everything was still evenly in tune, my guidepost began to really suck.

Long story short: I have no idea why it did what it did, but it was freaky. However, it was probably the most amenable side-effect I had from the medication.

-----

enf 100 days ago | link

Did tempos sound faster too, or did just the pitches change?

-----

ismarc 144 days ago | link | parent | on: How to design a class

When I first used oop, or any object orientation, it was in the early days of the public internet. There wasn't nearly the resources available now, especially for a hobbyist. So when I learned the syntax (I believe it was early c++), I had nothing else to go on. I naturally leapt to using objects to describe and build new data types. It made me think about what data I needed, then how the data needed to be operated on. It wasn't until I started taking courses that I heard the noun approach... And it just seems completely backwards. It isn't about modeling your data, it's about the conceptual items you're working on. I still think oop is a good approach if you ignore all of the 'best practices' and instead use it to build custom data types that simplify your algorithms. It's part of what I really like about Erlang records (even though they're a bit annoying to use) and Scala in general. You can build custom data types without having to just compose lists and maps but you get a lot of the benefits from functional programming.

I really don't understand why so much of the programming classes, tutorials, guides and common wisdom aren't 'data first' style design.

-----


Peppers are particularly difficult to cook with. Their oils are what is spicy, so you get variations based on growing conditions as well as how you prepare it. To reduce spiciness, remove all seeds, rinse with cold water and put it in a cold salt water bath for a couple of hours. Not enough to change the flavor, but just until you start to see the oils collect on the surface. This only works with raw peppers. If you then lightly sear the outside, you can get more of the flavor without the spiciness getting into the rest of the food as much. With the right sized pieces, you can then pull out the spicy chunks and still have the nice flavor.

-----

darkarmani 153 days ago | link

> To reduce spiciness, remove all seeds, rinse with cold water and put it in a cold salt water bath for a couple of hours.

This is popular opinion, but I've also heard from cooks that the seeds don't contain much spiciness in them, but the pulp around them does. Could it be that the outsides of the seeds are covered in oil and the insides are actually neutral?

-----


You should have a look to see how much memory the connection tracking table is actually taking up. IPTables stores a lot of its information in kernel space, but modifications are copied to user space, updated, then written back. As an example, for large project X with >100,000 users connecting through a linux-based gateway device, using a single firewall rule to allow access for each device grew larger than RAM available to the kernel. You can also tune the size of the connection tracking table (and pretty much everything else related), but 64k sessions was never a breaking point for us.

-----


Any particular advice on how to not lose data when map-reducing from a sharded collection into another sharded collection? Because we had to migrate away from mongodb in a hurry when we had ~20-40% of our data just not get written (basically 1-2 shards worth) from that scenario. And by in a hurry, I mean we spent a significant amount of time debugging, troubleshooting and tracking down what was causing the issue (we were on 2.2 at the time, not sure if it's been fixed since...) and realized there was no way to fix the bug unless we were willing to delve into how mongo did the writes from map-reduces. So once we found the underlying issue, we quickly migrated away.

The number of conditions where it will silently lose data and you have no control over the write consistency is absurd. However, every time it's brought up, people shout it down because they assume you're talking about the known (32-bit version and dataset size/ram size, not setting it to confirm the write, etc) write issues, and not completely different ones that aren't resolvable.

-----

manishsharan 168 days ago | link

@ismarc , I have never had to encounter your particular use case. Is there a bug report with details for duplicating this ? Like I said, I am not a fan of Mongo and were I to encounter an issue like yours in my use cases, I would bite the bullet and migrate to something else.

-----

ismarc 168 days ago | link

There may still be. The first two I opened were closed pointing to docs on how to set the writing stuff. The reproduction is pretty easy, if a shard tries to write its results to a shard that is write locked, none of that shard's map reduced data after that point is written to any shard. The more evenly distributed your data, and the more shards you have, the more likely you'll hit the condition. Combined with the fact that all unsharded collections always go to the same shard, the whole system becomes useless unless you can fit your entire dataset in all collections in ram on a single box.

-----


HUDs use light focused at infinity so the point of focus for your eyes doesn't change when looking at the display (it "hovers" mid-air). Holographic sites, HUDs, Google Glass all use the same principle, with varying ways of getting the light onto your eye (curved mirror that changes the focal point of the light properly that reflects on a piece of glass, a piece of glass curved to focus the light properly when it's projected on it and a lens on the projector to focus the light properly are the 3 primary approaches).

-----


I recently got back into regularly riding (I now regularly commute to/from work on my motorcycle) and I completely agree. However, I absolutely would love to have a HUD, but not anything people keep thinking you need a HUD for. A decent GPS unit + bluetooth or smartphone + bluetooth is good enough for those "I got lost, now need to get home" rides. I don't want any sort of communication provided on it. I want a real HUD, giving me real time, easily interpreted information about what's going on around me.

I envision something closer to aircraft/fighter aircraft HUDs:

  * Roll indicator
  * "target indicator" (in this case, any non-stationary object).  Can be as simple as a diamond shape near the center of the detected target
  * Visual indicator of direction/speed of the target (something like a line from the diamond as a 3d projection for where the diamond will be in 4 seconds based on current speed/direction)
  * Similar direction/speed indicator for you
If done with thin lines (like all the aircraft huds), all the same color, it presents a substantial amount of useful information in an easy to process/digest way without being overly distracting. But, definitely not an easy problem to solve, it's just what I'd like to see.

All of the HUDs I see coming out (Google glass included) seem geared towards providing you information to distract you from what you're doing, not enhance it, which is the opposite of what I want.

-----


The source that the quote is pulled from (Schiebinger, Londa (2001). Has Feminism Changed Science?. United States of America: Harvard University Press. ISBN 0674005449) includes (in the very next paragraph):

"In 1996 salaries for women in professional fields increased to 85–95 percent of men with similar jobs. Younger women in the United States (childless women between the ages of 27 and 33) earned nearly the same (98 percent) as men in their age group."

The quote from the National Science Foundation in the book is unsourced, however, the only related information from the National Science foundation I could find was http://www.nsf.gov/statistics/issuebrf/sib99352.htm which is not necessarily contradictor, but paints an entirely different picture.

While I disagree with who you are responding to, there has been a massive improvement in 1 1/2 generations and it removes a significant amount of credibility to present information in a way to try and say that widespread systemic discrimination still occurs in areas where it does not.

-----

quinnchr 246 days ago | link

The most recent information I can find says that the wage gap for similarly educated women in the physical sciences is 12%

http://blogs.nature.com/news/2011/08/gender_wage_gap_lower_i...

While I agree there have been drastic improvements, I don't think you can say structural discrimination no longer exists.

-----

More

Lists | RSS | Bookmarklet | Guidelines | FAQ | DMCA | News News | Feature Requests | Bugs | Y Combinator | Apply | Library

Search: