« first day (3071 days earlier)      last day (2242 days later) » 

00:02
Who's used a bit of PowerShell?
Trying to do a basic thing but my search-fu is weak
I wan... no, wait, clippy! I be doing a small script referencing a directory "I:\some\path"
When I try to use that dir with gci (Get-ChildItem) it complains that Get-ChildItem : Second path fragment must not be a drive or UNC name.
(I've used basic variable assignment -- $var = "I:\pathy\dealie" -- to set the path I want to use, but then Join-Path'd that with the filename)
Not sure of the propa' PowerShell way to go about it
ie do I do something with Set-Variable? Set-Location?
Well, sorted that by joining the path earlier on (thanks for being my rubber duck); now I just need to figure out why composite ain't working
00:42
Oh dear lord
Instead of using brackets, you have to use a variable with the bracket character
That took quite some searching
May be worth a self-answer question when I am more awake
01:11
It works \o/
stumbles blearily off to bed
01:42
Good job!
 
1 hour later…
02:46
@rahuldottech Occasionally, yes
03:38
For 2019, I resolve to be more courageous.
 
5 hours later…
08:28
booooooooooooooooooooooooo
morning
> I recently purchased, but haven't yet received, a laptop which has 4 GB of standard RAM, and 16 GB of Intel Optane memory. The laptop was marketed as having 20 GB of RAM, but after reading more about Optane memory, it seems that it serves more as a low-latency cache than as RAM.
booooooooooooooo
optane isn't ram
boooooooooooooooooooooooooooooooo
09:02
> Output 'adv_stream': Number of lagged frames due to rendering lag/stalls: 50221 (18.4%)
> Video stopped, number of skipped frames due to encoding lag: 45268/272720 (16.6%)
Muh frames! D:
That's... painful
09:24
@rahuldottech although if you use it as swap, it's probably decent -- DDR2-ish performance, maybe
@allquixotic still boooo
 
1 hour later…
10:26
I do love it when an answer is accepted, but not upvoted >_<
Makes me smile, or something.
And hats are gone?
One to keep an eye on
 
2 hours later…
12:01
Hello. What does IR-CUT mean in the context of this raspberry pi camera? (Sharing link below)
Does it mean it blocks infra red light?
Infrared cut-off filters, sometimes called IR filters or heat-absorbing filters, are designed to reflect or block mid-infrared wavelengths while passing visible light. They are often used in devices with bright incandescent light bulbs (such as slide and overhead projectors) to prevent unwanted heating. There are also filters which are used in solid state (CCD or CMOS) video cameras to block IR due to the high sensitivity of many camera sensors to near-infrared light. These filters typically have a blue hue to them as they also sometimes block some of the light from the longer red wavelengths....
If only tehre was some tool, which you could write random phrases into it and it'd give you possible meanings
If it really means that...then why are the IR lamps part of the product offering?
If IR is blocked = no night vision
@deostroll reading the product reviews - the filter is either dependant on light levels or controllable
@bob @JourneymanGeek @allquixotic theregister.co.uk/2019/01/02/wget_bug_fix
Is it me, or is this not really a bug :/
After all, the person DOING the wget.... already knows the URI.
@djsmiley2k Erm. It's storing the password ...
> the stored information survives being moved to a different filesystem, so someone wanting to steal stored URLs from can move it from the target's hard drive to a USB key with no trouble.
12:16
@DavidPostill on the local file system..... which you have access to
Hi sir, please send me the wget output of your accessing natwest.co.uk thank you
FIX: Format all usbs as FAT32
:D
@djsmiley2k It's still a security risk :)
Indeed
I feel like it's being blown up, as per usual
I have passwords stored on my laptop and a backup on a thumb drive. But they are encrypted :)
My bank now uses voice recognition if I call them ...
Because you're sane :)
Yeah, that's not at all scary right?
especially concidering they recorded you calling them....
@DavidPostill quick, stop storing audio in mp3 files as it leaks users voices!
@djsmiley2k They still ask some security questions as well, but the process is not the same as before the voice recognition.
I don't think any of the audio in my mp3 files sounds like me ...
13:01
(╯°□°)╯︵ ┻━┻
 
1 hour later…
14:14
Why can't folks just read the error messages they post?
You have answered your own question. Please read it again. "Does not seem like WSL enabled on this machine. Download a Linux distro from the Windows Store, run it at least once and then make sure to run in an admin powershell" — DavidPostill ♦ 1 min ago
And
Have you tried what is written in your question? This script is Deprecated. Instead use start-dfs.sh and start-yarn.shDavidPostill ♦ 13 secs ago
And
It tells you what do in the link you included. — DavidPostill ♦ 31 mins ago
@DavidPostill that's an advanced skill
lol
@JourneymanGeek Oddly enough they were all posted from IP addresses in the same sub-continent ...
I suppose some education systems don't encourage original thinking instead of rote learning (no offense intended to anyone who has been though such an education system)
@DavidPostill @rahuldottech would agree ;p
I absolutely would
;p
> 5 stars: Everything seems to be fine, but still something is missing... Heart is not fulfilled somehow.
#justIndianReviews
This dude keeps pronouncing LED as "lead" in this video and it's killing me
15:01
15:14
@rahuldottech as in the metal? Is it Dave of EEV Blog?
@JourneymanGeek Why is why I don't answer much anymore
I CANNOT BE NICE TO THESE NEEDFUL F*(^$WQSGDJ SNOWFLAKES AURGGGGH!
(╯°□°)╯︵ ┻━┻
@djsmiley2k Yeah, pronounces LED phonetically instead of as an abbreviation. Not Dave of EEV
Oh, he does too
now you know that, it'll annoy you then too :D
I don't know who Dave of EEV is
@djsmiley2k eh I answer some of those
It's easy rep a great chance to teach
I understand but my god
Also, no matter how many I answer, i don't get much (if any) rep.
They accept, and upvote someone esle ;D
 
2 hours later…
17:18
Heya has anyone used ext4's reserved blocks?
The use case is that a process needs to run, even if there's no space left on the root volumes
(if inodes are out, we are in a bit of trouble, but assuming that;s not the case)
One option is to pre-allocate a block, do a overlay-fs on the block to it's ext4 on ext4 (performance might sag)
Another might be the ext4 reserved block
@HackToHell ext4 automatically reserves 5% of your disk space for its own uses for this reason
idk you might be able to increase the reserved blocks space but it's not clear exactly what you're trying to do
all the ext4 reserved blocks do is they prevent ext4 itself from crashing when the disk is completely full
@allquixotic We have a agent which runs on devices(low powered devicies like raspi), the issue is that sometimes people run programs on it which kills the disk space in which case the agent dies. (it needs some space to run properly)
the journal can still be synced, etc.
@HackToHell that's an error in the agent code then
it needs to catch ENOSPC and EDQUOT errnos when calling write() (up and down the entire application stack and all dependencies and anything that could cause a write) and do something intelligent to stay running, but what "something intelligent" is, depends entirely upon what the agent is supposed to do
17:34
@allquixotic I know :( but it's large and will require too much resources to fix. So we are doing what we can and adding in hacks
Another horrible idea is have a overlay that writes to a file which is mounted as ext4
So ext4 on ext4 yay
@HackToHell does the agent need an unbounded amount of free disk space to operate, or a fixed amount? if the amount of disk space is predictable in advance, and guaranteed not to exceed a certain amount, you could set up user quotas to prevent them from driving the root filesystem too low on disk space, and run your agent as a user that has no quota
if users have root access to the system all bets are off; they can just eliminate the quotas
It's a fixed space, and users will have access to the root user
actually if users have root access to the system don't even try to solve this problem because any workarounds you put in place can be reversed by the user and they can then merrily go on and break the system by filling up the disk/RAM
OK, you're dealing with a human problem, not a technical problem
That is quite true
situation where: (1) you have a known defect in code and you're unwilling/unable to fix it, and (2) you have a human problem, stop trying to bandaid the problem, it isn't going to work
17:38
I can do that then, setup up quotas and tell people not to mess with them (add it in Documentation and hope they read it)
You will scream when you hear what the agent is :|
We are running it in a damn raspi of all things
humor me; what's the agent
you asked for it..
it's a salt minion
eh? saltstack?
Running on low powered devices :S
yup
if your users are at least a little bit / notionally cooperative with you, meaning they want the agent to run and want it to work, then the problem shifts most directly to the systems just being designed with insufficient disk space
especially if this happens often
17:43
There's the additional ambiguity with regard to this as users can use any linux board they want (os is limited to ubuntu for now)
@allquixotic -o loop -o size=blah boom
ah, so you don't control or provision the hardware
And the system itself is dynamic, the user can schedule applications on it. (we have had issues with application writing too many files and killing the device, it's the user problem most def, but if we can't have the control channel up(salt stack), we can't help them fix it either)
yeah, we offer recommendations but it stops there
creaTE A Loop block of the required size
presumiung it's not infinate
write to that
refuse to support if they don't have a big enough rootfs?
17:45
both you limit the agent to not go beserk, and you limit users not to kill it
but you can't code around dumb.
but yeah, quotas or a filesystem in a file is a bandaid, but users with a root shell can still go "you know what? this is taking up disk space, I don't want it" rm -rf /var/lib/salt or whatever
or killall -9 saltwhatever
@allquixotic the issue is even with a big enough rootfs they could break it. Just want to limit the probability of that
Most people using this aren't going to know systems so well
Which is sorta a selling point of our stack of sorts I suppose
but then that's business BS generator
@HackToHell you could ship a second process that stays running and monitors the salt process; if the salt process exits, create a FS in RAM and launch Salt into that (assuming it can work in the amount of RAM a raspi would have available)?
that would shift the burden to RAM albeit temporarily but give you your control plane back
@allquixotic that we could do, but with the fixed sizing requirements, just pre allocating a block should save me the trouble of writing one more process
Also we do try hard to not let customers use ras pis, but nope, it's cheap and it works :(
it's a bad business decision to offer support for a product on insufficient hardware; I would work with my management to convince them to eliminate support entitlements for anyone who runs the software on an extremely low-spec device like a raspi
you don't see Microsoft supporting Win10 on 512MB of RAM, do you? :P
it might boot but there's no way they want to support those configurations
basically they can download the software if they want, they can buy a license if they want
17:52
That is true, will have to work more on pushing back these deals.
but if they don't run with at least X storage (big enough that the chances of them filling it up are tiny), don't support them
And probably draft a better Service agreement to include these
you could also (to avoid a FS-on-FS situation) take control of the OS install process, meaning that your product would ship an OS in addition to whatever software you're offering on top of it - and your install script would just partition the disk to leave a sizable partition for salt and run salt in there
or require the user to partition if they insist on installing the OS themselves
We are going public this march, whew have to get this into management by then
@allquixotic This is exactly what resin does
oh wait they changed their name again
We are basically them + ROS based robots
also, amusingly, based on hearsay good wisdom from Bryan Cantrill, Linux and Linuxy programs are traditionally really, really awful at handling ENOMEM and ENOSPC (out of memory and out of disk space, respectively); both the kernel and low level system daemons have a bad time on Linux with these, but on say, Solaris and derivatives, and BSDs, these failure states are very robustly managed and the system keeps running
17:56
Interesting
I'm not sure I'd trust even systemd to work well in ENOSPC / ENOMEM conditions
but maybe Red Hat will surprise me
we've hit resource constraints with RHEL-based VMs in AWS for my day job
the fix is always to throw money into a larger instance type
trying to make the system sane with resource full constraints is insanity
And fs is just one part of it, have to deal with memory cpu and inodes
memory is going to be even worse
memory is probably set a low oom score and pray it doesn't get killed :|
oh man, lol, I need to find that talk Bryan Cantrill was giving where he was laughing at how ridiculous the OOM killer is
17:59
Please do!
to an old UNIX hand that's like designing a plane to set off bombs to explode itself when an engine fails :D
Also how you been @allquixotic it's been a loooong time
:D
it's been a while; I think we've both (?) been fairly inactive in here
I had a burst of high activity on Riot/Matrix and a bit in here for New Years but I've spent a lot of time doing hobby coding for my ESO guild
lots of learning Node and Lua: github.com/allquixotic?tab=repositories
but I still kneejerk into "this sucks, I'm using Java" when I really need to get things done and don't have patience to learn the quirks of the platform (ahem, Node)
Java isn't perfect but I understand its quirks at least
Nice, you too have fallen to the JS wave.
But then the current state of that ecosystem is impressive. I was recently looking at three.js and I was pretty mindblown by the state of things.
I like Node's ecosystem and even recent JS's expressiveness compared to Java, BUT... it continues to frustrate me with async and stuff like that
18:09
I haven't been all too active here either. Moved to my second job(and city) and been fairly busy getting the cloud side of things set up on OpenShift. I wrote some things in golang but haven't been programming as much as I would like. But running kubernetes has given me more insight to the linux kernel and network stack so yay.
await I would really rather be able to just go, sync () => { step1(); step2(); step3(); step4(); } instead of await step1(); await step2(); await step3(); await step4();
await openshift and kubernetes... huh.... I'm dealing a lot with Jenkins, Ansible, LAMP, Node/Puppeteer/CodeceptJS, and a lot of AWS thingos
await I'd like to be able to use kubernetes via EKS but we aren't that far yet
await I'm doing this await thing as a joke because that's how I end up programming JS 99% of the time :P
:D Yeah the complexity of kubernetes is staggering. The network stack is just terribly hard to debug.
But then they do simplify a lot of things for the end user I guess
EKS makes the networking simpler to configure
I personally prefer LXD still but idk how well that integrates with enterprisey management crap like openstack/openshift or other control planes
Unfortunately there is no managed openshift yet.
If there's a CNI interface for it, it should be easy
actually my favorite thing ever is Solaris Zones
18:13
The number of docker clones is staggering
Have you been looking at cilium and bpf
Looks pretty exiciting
@HackToHell RHEL 8 is shipping with Podman, which is a daemonless docker clone that supports the same CLI options
so literally a docker clone
@HackToHell not yet
@allquixotic ah interesting. Redhat recently aquired the CoreOS team as well, and they have rkt if I am not mistaken
cilium allows you to add l7 firewall rules to the damn kernel
lol. are we now avoiding retpoline performance overhead by running everything in the kernel? :D
lol
I have lost track of the number of fixes that have come out for it
@Bob btw, you were asking about a simple bidi TCP-based proto? grpc.io
18:22
Most CNI implementations run damn BGP for keeping track of the pods.
18:54
woo, I'm writing new code that is vulnerable to the y2k38 bug :D
19:07
lol
Is time64_t available on your platform?
(or some other 64-bit time representation)
19:46
Soooo the nono car is gone
20:24
@djsmiley2k nono car?
!!nononocat
@allquixotic That didn't make much sense. Use the !!/help command to learn more.
!!listcommands
@allquixotic help, afk, awsm, ban, color, convert, define, die, doge, domain, eval, export, findcommand, forget, forgetseen, github, google, hang, imdb, import, info, inhistory, jquery, learn, listcommands, listen, live, mdn, meme, moustache, mustache, norris, nudge, parse, refresh, spec, stat, stats, tell, timer, todo, unban, undo, unonebox, user, weather, wiki, xkcd, youtube, zalgo, !!!!!!!!!!!!!!!!!!!!!!!!!!!!!, 65wat, ;), ;p, \/s*, adminspotting, after5, ahh, ahhfire, apumpkin
areyoukidding, baaaaat, bababababat, bankerror, bankey, baroo, bearhello, beatingbloodoutofarockwithahalberd, b
!!nonononono
20:26
yowl
 
1 hour later…
21:27
Wow. I'm using Java Contexts and Dependency Injection (CDI) 2.0 with Weld, the reference implementation of the standard, using Java SE (meaning, no application server). It's only a matter of bringing in a dependency via Maven, writing a short XML file, then using two annotations.
good intro to DI while using a standard/stable/tested DI framework. I tried to use Dagger2, a DI framework from Google that's "lighter" (no reflection, so very little runtime impact), but it screws with Lombok. Weld is runtime. It adds about 2 seconds to startup.
Bob
Bob
21:51
@allquixotic yea I ended up using websockets
except now I have to fix a bug stopping the server from shutting down
goddammit java
thread synchronisation is such a shitshow
@djsmiley2k you're reading the register, of course it feels like it's being blown up.
my first project with successful dependency injection :) github.com/allquixotic/nannybot
 
2 hours later…
23:38
Lol true.

« first day (3071 days earlier)      last day (2242 days later) »