VirMach
I think there may be enough to say about virmach to warrant a thread unto itself - so may as well start this one right here right now 
(Following some recent discussion in another thread here)
For now I'll just summarize my assessment from last couple years in a few of their USA locations: solid systems, great uptime, modest specs, okay-ish network (depending on what you need it for - it is colocrossing after all) ... Also reasonable to deal with in terms of support.
So all things considered I've been pleasantly surprised. I'm guessing that the CPU limits (etc) are actually fine for some medium-load webserver duty (such as this forum) - but probably less so for more sustained usage patterns such as transcoding.
tl;dr: could be a good option for many but not all lowend applications - and proven to be worthy of consideration (in my own limited experience at least).
Questions, suggestions, critical comments, horror stories, and/or kudos and benchmarks welcome.
Also feel free to mention any good alternatives in the general price range
HS4LIFE (+ (* 3 4) (* 5 6))
Comments
Apologies for not including more gory details in the OP. I'm typing with my thumbs, 6 hours into an 11-hour train ride ... Will be adding more bits and pieces in subsequent posts in this thread at random intervals - possibly for the next several years. (That's an inside joke referencing the 2018 BF flash sale thread from the old forum ...)
HS4LIFE (+ (* 3 4) (* 5 6))
Take a nap; sober up. Godspeed!
"It's a hard life- to be a stick insect." - Karl Pilkington
"The good stuff comes on slow ..."
HS4LIFE (+ (* 3 4) (* 5 6))
Also, just to note that VirMach is not exclusively CC, e.g., at their Seattle location.
Their services are rock solid. But, if you are thinking about using their promo services purchased during BF / CM / XMAS for anything beyond initial bench-marking, idling and hosting odd websites or services that require sustained Disk I/O and 'fair share' of CPU cycles, you are better off looking elsewhere, for you'll likely get a crash course in learning to manage expectations and a resounding reminder that you get exactly what you pay for :P
I'm using their BF specials for sustained 40 iops, which isn't 'heavy' but also isn't zero. They clearly publish their limits, and I'm in a position where I can throttle my own stuff to stay within them. The most annoying thing for me is that they hit me for iops once, and gave me a log of it, and their log numbers were significantly different from what I was measuring in
/proc/diskstats. I don't understand why there's a difference or where it came from, so I ended up just scaling my measurements by a factor, which is an ugly hack.the best thing about Virmach is rock solid stability. IMHO its really suitable for production but in order to do so i believe the resources are really limited on the CPU and I/O side. perhaps such is the specs for BF deals. A simple nench shows some insight on how it is regulated.
I bench YABS 24/7/365 unless it's a leap year.
The same happened with me once.
While I still keep a couple of $3/y and $5/y services with them that I use for playing around with random scripts, apps etc, I have largely migrated casual projects and personal web services to other hosting companies that do not race to the bottom with their prices and consequently do not suffer from overcrowded nodes or general paucity of resources. Also, servers in USA have little use for me, so that has certainly helped me move on from some seemingly incredible deals I had with Virmach.
At the end of the day, as long as one understands his/her requirements vis-à-vis limitations of said VM, things will likely end on a pleasant note.
I had some question about their KVM lite.
I had OpenVZ with them which was recently migrated to KVM lite for free.(bought it in last yr sale)
I dont find any AUP for their KVM lite. Does someone know about their AUP.
What is the cpu share for KVM lite.?
50GB SSD
3 cores
2TB BW
2GB RAM
12$/yr.
KVM
https://webhorizon.net
I've got 2 VPS that haven't provisioned yet, one since Friday. Presumably other folks are the same?
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
Funny how we moved to a new forum but have no problem advertising CCs biggest progeny and stating "it's super awesome".
I mean, I get it, a stable cheap service, is a stable cheap service, however, you have to see the irony in this thread being here (regardless of the fact they 'use more datacenters than only CC'). Their main marketing point and ability to offer said services so cheaply is the fact they rely on CC to basically subsidize them with cheap servers and hosting like any other CC elevated brand has in the past.
Just saying, moving to a new forum because we are wanting to get away from CC, then we advertise their largest brand here.. for sure gives me a chuckle to read...
Cheers!
Have an Allwinner H3 device? Android? Check out H3Droid! | Lichee Pi Zero - The 6$ SBC | #SYSarm - Get It! | Atomic Pi - $35 x86 SBC
20+ Years IT Experience in Linux/Windows Hosting, Administration and Development
@TheLinuxBug
Do you often cut off your nose to spite your face? Just saying. ;-)
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
If your question is, do I avoid all CC properties and related providers like the plague? Then the answer is most assuredly, YES!
So, yes, in this case I would, because, I personally would NEVER recommend a single provider mixed up with CC to anyone.
That's just me though, for some people price is more important than anything else, so I am sure they will blindly continue to purchase anyway. I just won't be one of those out there promoting them and suggesting people gamble on it.
my 2 cents.
Cheers!
Have an Allwinner H3 device? Android? Check out H3Droid! | Lichee Pi Zero - The 6$ SBC | #SYSarm - Get It! | Atomic Pi - $35 x86 SBC
20+ Years IT Experience in Linux/Windows Hosting, Administration and Development
hello @Abdulla
I would just assume resource limits as described in https://virmach.com/terms-conditions/#tab1
(these do seem sufficiently restrictive
)
HS4LIFE (+ (* 3 4) (* 5 6))
@TheLinuxBug
it's a fair cop.
I am indeed an incorrigible troll.
HS4LIFE (+ (* 3 4) (* 5 6))
Drat! Missed the $1 - OoS
:'(
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
I didn't, but shit still died before I could checkout.
"It's a hard life- to be a stick insect." - Karl Pilkington
Well, it did die on me 1st time but when retried it was OoS.
Got a nice Cloudflare page after that.
Ach, maybe there were only 5 available
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
Thanks for reminding me about the history of virmach.
Cheap as they are, will not be supporting them any longer.
I bench YABS 24/7/365 unless it's a leap year.
Just learned about them now, after renting for a full year. At least I'll pass on the $50 8GB/80GB from them. Might get UltraVPS 8GB instead, even if it's higher pricing. Their 4GB SAS is the most reliable VPS I've ever had and the pricing is still unbeatable for the reliability.
Doesn't AlphaVPS use CC for their LA location? So you'd put Alex on your blacklist, too, even his UK/Clouvider location?
Mine said it was "Active" but also said something like "Virtual server not found" on the Manage Product page. If you go to the Control Panel the server was there and working fine.
I had that until late last night - now Active and Solus available.
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
So the next question is, can I use HE.net or tunnelbroker to automagic a proper IPv6 address for my CC Virmach?
Thank you. Yes it is restrictive enough.
https://webhorizon.net
That makes Any Low End Spirit box look Prettier
Replying to myself. Yes, easily, and the speed is pretty decent (so far).
Just send them a friendly note that you did this. The TCP fixup notices spam the shit out of the hypervisor kernel - and it can't be turned off. If they don't set it to ignore these notices, you will be spamming the living fuck out of the logs.
"It's a hard life- to be a stick insect." - Karl Pilkington
Will do, thanks, I had no idea!
Most don't. It's annoying for admins to deal with, and makes work a little more difficult if they're not already setup to handle this.
That said, I ran n+1 (where N=last) as a backup using HE+VirMach for a couple years in the event of everything dying outside of Buffalo.
Worked great. Just a bit slow for UDP and a shitty way to add IPv6 to the box.
"It's a hard life- to be a stick insect." - Karl Pilkington
Never used them because they are CC clients. Same applies to every other host which is a CC client.
Recommend: SmallWeb|BuyVM|Linode|RamNode
I played nice and disabled it and asked for permission. I'm getting old in my old age.
Now I have two to sell. :-/
In stasis until the shitposting stops/abates.
Than=compare;then=sequence:brought=bring;bought=buy:staffs=pile of sticks:informations/infos=no plural.
It wisnae me! A big boy done it and ran away. || NVMe2G for life! until death (the end is nigh).
this might be a good time to point out that VirMach uses ColoCrossing for almost all of their locations ... (just in case anybody might have missed that important detail).
While I'm inclined to believe that they are not a summerhost or typical CC scam/front company, it may be fair to assume that they do have a significant relationship with CC - and I understand that many people here would like to draw a hard line to make a clean break from the ugly BS coming out of CC for however many years - that being one of the things that lead to this very forum rising up from the ethical hellscape of the old one ...
So I do respect and welcome the comments such as for instance the most recent one from @vpsgeek above. I'm not going to pretend to know so much about the bigger picture, and am just sort of following my gut here (maybe not unlike someone setting up a McDonalds at a vegan cooking teach-in) ... I guess I'd be most interested to read more informed comments to get better insights with regard to VirMach's business practices and whatever philosophical perspective on the ethics of doing business with them, if there is any more nuanced analysis that might emerge. Or not. I'm okay with the commentary, just would prefer that it doesn't get too hyperbolic or repetitive. But Informative or insightful or even cathartically hilarious digs are all good in my book, so please do feel free to chime in if that's what's on your mind.
Anyhoo ... I've been camping on their flash sale page via @FAT32's excellent https://virmach19.fat32.top site. And finally bit on a $48.60 per year deal in Seattle (which I understand may not actually be with CC? Bonus! But still no IPv6.)
I think this was one of their better flash sale deals this year - for my needs at least.
$48.60 got me 10 GB ram / 4 vcores / 30 GB ssd - and SolusVM control panel which made for an easy install of Debian 9 via ISO - no glitches in that process.
The ssd has so far been fairly consistent for sequential write speed in a range between 120 MB/s and 160 MB/s (I am logging a 256
ddevery 10 minutes over a 24 hour period to see if any downward trend as the node gets loaded up with new users)CPU is generic QEMU clocked at 2.1 GHz with very few useful flags enabled.
geekbench 4: https://browser.geekbench.com/v4/cpu/14992890
if I'm remembering correctly, I noticed up to 20% cpu steal when running the multi-core part of the bench. Will be interested to see if that is a persistent condition on this node.
I'm going to idle it for a few days, then start messing around with the system in earnest. Probably will let the dust settle for a couple weeks before checking in with a low-priority support ticket to inquire about enabling
aesflag to facilitate full-disk encryption and ssh connections.And that's where I'm at with it so far - about as expected, getting pretty much what I paid for: a (presumably) stable system with a lot of RAM and generally adequate low-load disk and CPU in a west coast USA location for less than $50 per year.
HS4LIFE (+ (* 3 4) (* 5 6))
How about the other way around? Ipv6 to ipv4
I bench YABS 24/7/365 unless it's a leap year.
Just so you know, you'll eventually get asked to stop it, or get turned off. That still impacts others, and once the smoke clears, smackdowns for overuse will be noticed on a per-VM basis. Since there's no non-realtime way to monitor your use, I'd figure you were probably torrenting, or lying about running benchmarks every 10 minutes (who the fuck does that?) and bitch you out for it.
That said, I do not represent Virmach.
"It's a hard life- to be a stick insect." - Karl Pilkington
I think they don't allocate full threads but who knows
I bench YABS 24/7/365 unless it's a leap year.
hmmm ... well, alrighty then - thanks for that perspective (I do appreciate it!)
While I have yet to get any complaints from any of the dozen or so providers I may or may not have impacted with this extended intermittent disk test - I will be mindful to check in with support before doing these tests in the future. And - thinking about it some more now - it may make good sense to scale back the size of most of the disk writes to less than 256 MB for the intermittent samples, and maybe also add just a couple larger writes (something like 1 or 2 GB) to see if the filesystem / disk caching is a significant factor.
I do have some vague awareness of the intricacies of benchmarking disk performance - this is certainly one of the simplest (and possibly lamest) things I know how to do to get a rough indication to compare on the different systems I'm evaluating.
For what it's worth - my reasoning for considering this as a relatively gentle benchmark is roughly as follows: cumulative activity is about 1.5 GB per hour - generally taking much less than 3 seconds out of every 600, so under 0.5% utilization - if time spent is any reasonable measure (which it admittedly may not be) ... for a total of 36 GB written in a 24 hour period. Which in this case is a bit more than the 30 GB SSD size .... but a small fraction of the 7 TB bandwich allocation for the month. Not that I want to conflate disk i/o with network activity, just taking into account the relative scale of the activity.
I guess I'd be more concerned if I were doing the disk write tests 24/7 rather than a few times a year. But you're right - it may be pushing acceptable limits. Anyway, some good food for thought and motivation for a less intensive approach going forward - thank you for sharing that insight.
HS4LIFE (+ (* 3 4) (* 5 6))
As well you should. Just keep in mind that they're one of the cheapest options, and I have no idea how they make money.. but before they refined their monitors, I got a nastygram for doing an OS install reformat. It's nowhere near that level now, but common hard writes with implicit flushing several times an hour do affect other users.
"It's a hard life- to be a stick insect." - Karl Pilkington
indeed and indubitably - I realize now these day-long assays have been sloppy and gratuitous.
I guess a rule of thumb to estimate impact on the node is simply to consider the possibility of 100 (or more) other users doing the exact same thing.
As an amateur expert in the Dunder-Mifflin effect ... I find myself now beset with much doubt, and rendered (temporarily) insensate.
And I thank you for that.
HS4LIFE (+ (* 3 4) (* 5 6))
Lol, it's wednesday and cyber monday flash sale is still going on. I've had a few of these (thanks WSS) and they have been solid, though cpu speed hasn't been the best. I'm down to just one now, running a lightweight service, so performance isn't an issue and reliability has been great.
Maybe the bot revolted and echoed
Bitch I run this show !
I think they didn't sell enough. Think most of them should be 2660s
I bench YABS 24/7/365 unless it's a leap year.
You mean creating an IPv4 address on IPv6 only VPS? No idea. Probably not.
I wonder if virmach is impound yard of all those deadpooled CC summerhosts' server gear?
boooriiing!!
Hi!
If anyone's looking for some fresh hell of the San José flash deal persuasion ... https://virmach19.fat32.top would be a real good place to be right about now
HS4LIFE (+ (* 3 4) (* 5 6))
Crap, I could have used some of that. There's some more up now but the current one isn't what I want. Will keep watching. Should set up some way to monitor the bot and alert on istuff of interest.
@willie lol the party is just getting started ... the billing server has apparently caught fire a few times,. captchas and 502s are flowing like cheap wine down the drain on the way to some putrid sewer to slake the thirst of 10,000 PRC zombie bots pitching a fit ....
To mix metaphors a bit more, imagine a fireworks display fueled mainly by low-grade potassium and amyl nitrate, punctuated by occasional bursts of high explosive.
tl;dr - the finale is when it gets interesting - but it can be a wallet-draining guessing game until the porky pig gif appears if you stumble into the action without a full appreciation of that FOMO dynamic
anyhoo - what specs are you looking for? I'll try to keep an eye out for something suitable while I'm dodging the buffaloe stampede
HS4LIFE (+ (* 3 4) (* 5 6))