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
Thanks. I was 0.1% worried.
This seems the most likely.
I moved my VM from Montreal to Toronto. I don't think it has anything to do with network settings or other magic. I haven't touched anything like that.
I've never had any issues with hitting 100+ mbps upload with Verizon and sustaining it. But maybe it's the connection to Cogent.
As for everything else, the only thing I think might be remotely the issue is the CPU with SSH. Even then, I'm pushing ~13-14 MB/s right now to the server (in bursts, not sustained), and only then is the CPU hitting 100% load. It's like 20% load at ~30 mbps. I have 1 core with Lunanode.
Yes, I rebooted it from the get-go once I suspected something was an issue.
If your CPU can't handle SSH without puking, congraulations- you're on a VAXstation!
That said, you've endured the huge amount of shitposts all over your thread and not even bothered to react. Good for you!
If you're pushing 100% CPU with 13Mbps, you're probably on a HDD node with a dead
lockline scheduler. Not much else to do but ask them to deal with noisy neighbors, deal with it, or move on.I didn't have any issues in Montreal. I just noticed that you guys recommended Toronto because there were more Ceph nodes than Montreal and Roubaix. Oh, and you had the deal recently, so I used it as an opportunity to move my volume to Toronto. I can give you my IP / open a ticket.
Otherwise, I am getting ~14 MB/s uploading to the server off and on. Sometimes it drops back to like 2 MB/s, though.
Let me check the CPU model / speed. But yes, I saw it hit 99% in htop at ~120 mbps (14 MB/s), which is my max upload anyway.
Toronto:
Toronto openssl speed rsa:
Montreal:
Montreal openssl speed rsa:
I'm sitting at 37-55% cpu in Montreal while I upload to it at 14 MB/s. Note: Toronto is Ubuntu 16 I think. Montreal is a fresh install of Ubuntu 18. As you can see, I the tests are slightly different (Montreal goes up to 15360 bits for the tests).
Some one rang the Yeti Assassins?
I just tried a 4-core VM in Toronto, and CPU definitely should not have been an issue. I am getting similar network speeds with it. I'm getting below 20 mbps (2 MB/s) upload to it at times.
There's something unsettling going on. Montreal, unless hosted on a 10 year old laptop, has a different CPU/hardware model, and I can't think of any specific reason why other than to artificially limit things.
You keep posting and posting again, but still couldn't simply run the test and show the results.
Maybe indeed like the first reply said, you're not interested in diagnosing the issue, but just lonely and want to talk.
I'm not sure what your test is going to prove. Lunanode / perennate already said on this thread it's probably something between my ISP and Cogent. I created a ticket, and they saw some packet loss that could be causing some throttling.
I ran the script, and it's been running for 2+ hours. The speedtests don't take into account a really slow connection to one of those servers, so when the connection is 200 KB/s it takes forever to download 100 MB. I'll run another script if you send me something that doesn't involve a hard-coded speedtest.
@cirrus_cloud if you ever decide to read an intro book on networking; you will get an answer.
OK.
If they did, then perhaps it can switch that from probably to certainly.
You can try this one: https://bench.sh/, it doesn't test as many of the Asian locations, which could have slow speeds for you.
I'll edit the other one and will remove the Singapore and Vilnius locations that were taking a long time.
I got rid of all of the really slow locations (some were like 200 KB/s):
@cirrus_cloud thanks muchly for doing the needful ...
Noticing OVH and Softlayer are at the slow end of this pool - which does point to peering issues with cogent (which are likely due to cogent's business decisions and maybe not so likely to get resolved anytime soon, but who knows)
I would be interested to see list of the ultra-slow ones you removed from the script altogether
So it looks mostly just fine to me. If you would also not fuck up the formatting entirely, and just post it with the <prе> </prе> tags.
Ah yeah.
That's exactly how benchmarking works.
Let's get rid of all the low numbers.
All good.
I'll run it tonight with the slow ones included. Singapore, Japan, Australia, and one in Lithuania I think.
That is quite the list of speedtests ... seems like most are fine from Toronto except a few which are really slow.
Might be more useful to just do an
mtr
for the slow ones, no need to torture them with more speedtests unless somethings changed to test again. (Not suggesting you actually do any more unless it's interesting to you.)I did notice much better speed (full 100 mbits) to kimsufi in BHS last time I checked.
I'm guessing the problem in general has to do with cogent's peering (business decisions), and the solution you've found is to use Montreal instead of Toronto. (I'm okay with the "good enough" cogent in Toronto so far, but am not currently using it enough to really notice problems if any.)
I decided to run a test.
Lunanode vs Hivelocity vs Virmach vs Hosthatch
Looks fine.
My nuts itch. @3606202
This is all of them. There are no 200 KB/s tests on here, so it's definitely better than when I first tried.
Interesting.
I thought you are female.
That's $30 extra.
You don't even have a provider tag.
https://www.lowendtalk.com/discussion/153228/rules-for-selling-on-lowendtalk-updated-04th-jan-2019/p1
The 30 bucks is just to cover processor fees; the service itself is free.