New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Yeah looks like some people got vswap.
Fixing that up now
Francisco
OMG Fran will kill me now
@yomero I doubt it, I had my load over 2000 before my shell froze up :P
I broke vswap fully for new provisions.
Damn thing.
Francisco
Past 10,000 procs inside of a CT seems to be the breaking point.
I'm not sure how but Hien was getting up to like 12,000 in a forkbomb, where as everyone else is only getting to 2k - 5k tops.
Francisco
I've got me a test box! I'll do my best to mess things up for you
I have got my trial account and it says active but still no server details.
Order Number: 1943768021. Look forward to trying it out for you guys
load average: 28.85, 12.40, 4.89... doin it right? not quite. I'll keep working on it.
Yep, mine reads active as well but no service. Order 4166493180
Ok, I am going nuts xD
And growing
Oh my god yomero. Such awesome stats xD
@yomero please do dual prime95 and pi-test.. and maybe you can put folding @ home too. :P
LOL!
But that just will get me a load of... 16 maybe u_u
Edit: My last one, now I am stuck in the shell
So, I have more than 10k processes, my shell is stuck, but I am alive afte one hour or so.
Alright.
Well, Anthony locked down everyones max processes because they won't stop forkbombing things like mad. It isn't really helping test the nodes :P
Francisco
u_u
At the end I am kicked XD
Hmmm, I think you must instruct us in how to do it
geekbench, unixbench, prime95, loadimpact, etc.
At this point we're happy with the results minus a few adjustments needed on stallion related to vswap.
Francisco
what about
running at the same time (2 ssh window opened)?
bold : change it to whatever you want (can?). :P
You can if you won't but I'm not sure how much that'll help benchmark it :P
It's proven at this point that I/O strain doesn't affect this kernel at all. Really, at this point we're just letting the trials go on for fun, not that we really need it to debug anymore.
Francisco
Hahahaha nice.. i'm more interested to go to your cabinet and pull the power plug out. now that's more funny. :P
ps: have a stock for storage vps? i'm still waiting. :-w
I don't think we got any storage spare unless @dimecadmium cancels his like he keeps stating :P
People are holding onto their storage plans for dear life
Francisco
yeah..
Hmm om, moved to hosting stuff
A single wordpress page of 1.5MB getting requested with apache bench very hard xD
And don't do I/O stuff please u_u. Sometimes it get sluggish
It gets sluggish because there's:
The box is getting ripped apart but handles well minus the massive forkbombs. OVZ documents that past a certain point things go down hill fast with OVZ. It's obvious to us that .32 is just threading much faster than .18. On .18 the few CT's we've seen get forkbombed, they only get to 500 - 700 processes before they simply can't push higher.
Francisco
*another user has an infinite loop spawning ffmpeg instances which are encoding a video file into individual jpeg images
... or should I say had :P
Me yesterday xD
Please use it without the I/O tests u_u (btw I don't know how, someone can tell me? )
I did that before but not anymore. BTW. Are we meant to have any vSwap, apparently I don't.
Vswap is disabled since some days ago
vswap is the devil and breaks things pretty horribly. They might have it fixed in a newer kernel build, I won't know until the devs get the next roll out done. When they get that, we'll beta it internally then roll it to the test nodes
We'd like it if vswap was working correctly and not needing us to likely introduce ghetto-burstram, but we'll see.
Francisco
@yomero tell me how to spawn 10k processes, i will contriboot