All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Being a good customer
I have two VPSs through ChicagoVPS (who are very good, would recommend to anyone) both hosted in different locations. Lately I've had a bunch of slowness on one of them, but its been very sporadic. It'll be slow for a little while, and then magically just fine. If I run this test:
dd if=/dev/zero of=test bs=64k count=16k conv=fdatasync;unlink test
I get mixed results. It'll be 17.7 MB/s, 22.5 MB/s, even 7.7 MB/s, but then it jumps back up to the normal >100MB/s.
I submitted a ticket, and at the time they ran their test on the main node, it didn't report slow. My question(s) is/are this: what can I do to make sure that (a) it's not something on my VPS that I'm doing that's slowing things down and (b) that I'm not being a pesky annoying customer saying "oh my node is slow, I've tested it and its slow fix it"?
Comments
Well I think the most direct and beneficial way for both of you to resolve it is to request that they enter your VPS container and reproduce the results and can troubleshoot it from there.
Please also be prepared to pay an admin fee if it turns out to be some error on your part.
But that gives everyone what they need to do full end to end troubleshooting and the fastest resolution.
Guess your vps is openvz based, and as there are other users on the same host as you, I think what you found is quite normal; say some other user might be running some i/o intensive applications as you were doing the dd tests
They can monitor for disk abuse but if your testing just happens to collide with a reasonable quick burst from someone else then that's just the nature of OpenVZ. Honestly I don't think CVPS oversells as much as people think they do. They do allow game servers and on SATA disks that's a wildcard at best.
I'd run the test once every hour for 2-3 days and compile a full spreadsheet for them. One burst every hour for that little dd test shouldn't be a big deal.
Yup, OpenVZ.
It's not so much a complaint on the service, more a departure from what I've experienced in the last few months. The last time I had reported it, there had been an abusive user that was dealt with pretty swiftly.
I'm going to look at cron-ing out a job to get values once an hour or every 2 possibly if it keeps happening. The worst is when its out long enough (5 minutes) to trigger a "down" notification from Observium. It still pings, but Observium checks SNMPv3, so it needs a response from an actual application as opposed to the network subsystem.
If its choking to the point you can't process data, big problem. I know they're working on doing better, so @CVPS_Kevin get some active monitoring set up with threshold alerts
That can be considered abuse. We had users doing serverbear tests every hour or so by cron and that meant termination.
dd test should take under 10s in a bad case. Perfectly reasonable unless the drives are constantly over utilized.
Could do other tests but dd test is what triggers people's brains most commonly, love or hate it. They should still perform adequately.
@jarland I would rather not have someone burning out my drives, doesn't matter if it's one person vs ten doing it per hour it will still decrease performance.
There is also this unfortunate thing that they put full hour in the cron. This will coincide with other cron's at the full hour.
I didnt see it to be a problem so far, only on the very big servers slightly, but in theory it may happen.
Sign up with provider that undersell.
You mean like Prometeus, I came to see it somewhere in this regard.
yes, and many others. There are some in US. Sorry, I meant under-load not under-sell.
That and nodes with better performance, i.e. better and faster hardware too.
Yes. Maybe choose SSD based, or SSD_cached offers.
but that's your right? isn't it? wouldn't it be annoying to you if its frequently slow?
It might be my right, but is it the best thing to do in this case? Should I be doing more to make sure I'm not the cause?
I think you need to make sure you are not the cause. You might get terminated if you are.
edit:
I mean sometimes it happens that you mis-configured something and causing high loads that you did not intend to.
If its impacting the node your provider should be able to tell you if you're the cause.
Ideally should not need a ticket for that either.
There are many quality hosts here at very affordable prices.
Some have a lot of locations to choose from, others have big traffic, others super-speedy server, in EU or US and even in asia and south america you can still find affordable VPS hosting from serious providers.
Not sure what the problem was with my suggestion however.... http://www.lowendbox.com/blog/how-to-tell-your-openvz-vps-is-swapping/
Take a look at that.
VPS = shared resources. Hence, you cant get all you want all the time.
If anything, there is no 'slowness' at all. You haven't paid for a guarantee, just a share of what is available.