All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
OpenVZ VPS, bad providers, and services down, what's the best act?
Hi,
In OpenVZ it's happening (like I've got right now while writing this), that a provider proves to be crappy after sometime of getting the VPS, or sometimes oversells right from the start.
An example is my provider does a lot of overselling. With using the new OpenVZ memory allocation (or swap) or the old allocation type (burst), it's still the same.
The problem that happened for me is finding at times, an individual service/s is dead every now and then.
Sometimes it's the web server, database server, or DNS, even a VPN or proxy death. And I come back to find the whole website is down for example.
That happened with me in yearly budget plans, but I've seen some reporting similar memory, responsiveness, or service run problems. Even while they themselves might not be even close to 50% of their allocated RAM.
What is the best practice once as described a problem is looking to be totally out of my own container (ie overselling)?
Is it for example to run a script that checks regularly all required services and if any is down it runs a command to start it? And if so do you have any good example, so it can be used to lessen the headache of those crappy providers?
Thanks!
Comments
It's about (mainly) RAM overselling in OpenVZ, where your services dies for lack of memory even if you yourself isn't using 30% or 50% of your VPS total RAM, but because of overselling.
In such case is it best to quit/move, or to do some solution? (it's not happening often for me, but once in a few days or in over a week).
If the provider has a good reputation, ask them what the problem is.
In your case, you state:
So your question boils down to "I have a crappy provider - should I stay or should I leave?"
It's not always overselling, it could also be from port scanning / brute forcing (i.e. 10 bots connecting to your sshd or to your imapd and this way exhausting all your memory and then the next process that happens to request an allocation fails).
Now about the cure - maybe setup some external monitoring that periodically check if the website is operational and if it's not - sends you an alert. Or automatically reboots the VPS somehow.
To be frank I'm not used to opening lots of tickets with unmanaged, so usually I save it for node down and real issues. But not for QOS issues unless it's becoming extremely bad..
But I might eventually contact them and see what they answer.
It's not always overselling, it could also be from port scanning / brute forcing (i.e. 10 bots connecting to your sshd or to your imapd and this way exhausting all your memory and then the next process that happens to request an allocation fails).
Now about the cure - maybe setup some external monitoring that periodically check if the website is operational and if it's not - sends you an alert. Or automatically reboots the VPS somehow.
I thought of a local bash script that checks and keeps required services up and could be without reporting back to me.
Local bash script is not the best solution, because it can die too, the same way as other services die. The monitoring must be remote.
Ok and won't disclose it for the meantime
If you haven't paid yearly, I see no reason to stay. It all depends though, the company could be okay, but the node could have some abusers on it. Some of the better companies around here will monitor their nodes, but if you don't let the host know about the problem, they're going to think everything's fine.
I've got monitoring for the whole server, and port monitoring but not applied though.
Could it be a local script that also listens on a port and can be monitored with a remote monitoring service that notifies by sms/email just when the script itself is down?
Edit: Does it effect cron jobs as well? As the script doesn't have to run on the background but just with a cron job.
Of course, there are many ways to skin a cat Do whatever is easier for you.
It's yearly but not too pricey, and used it for sometime so the lose is just so small, I'll try contacting them though.
I'll better be searching first for an available similar project that checks and keeps services up, as I'm not so good to write one in bash.
Always best to contact the provider if it happens for more than a few hours.
There may not be much monitoring on the VPS nodes and you could help them solve a little problem before it becomes a major problem!