All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Noisy neighbor or was it me?
I am trying to understand something, as I have never encountered this before. I have a VPS from a low-end provider (OpenVZ, in this case). Yesterday, I got an email alert from lfd (I have CSF installed on the VPS). It read as follows:
Time: Sun Mar 23 20:11:01 2014 -0400 1 Min Load Avg: 20.17 5 Min Load Avg: 7.79 15 Min Load Avg: 2.99 Running/Total Processes: 2/91
I checked, and the VPS had been restarted, per the provider's policy, I suspect. The weird thing is that the only thing on the server was my Virtualmin install (with associated services, of course) and CSF. If I am evaluating a host, I like to check uptime, how well the host communicates to its customers and performance before hosting anything there. So, I try to keep things minimal while I am checking them out.
So, my question is, could abuse by someone else or a crowded setup be responsible for what I saw? While my suspicions say yes, I can honestly say that this is a first for me in my many years as a VPS consumer. Have I just been lucky all this time or should I be looking at my own setup?
Comments
As far as I'm aware, no matter how noisy your neighbours are, it'll never reflect in your load averages in OpenVZ. So I suspect you have a rogue process in there somewhere - or maybe a rogue PHP script etc.?
I would disagree. I've seen a load of 10 on a OpenVZ container, when I checked top I saw none of my process with a load higher than 0.01. I confirmed with the provider that I had an abusive neighbor.
We noticed that this happens when the node gets hit with a large DDoS.
Load is calculated based on the number of running process plus the number of runnable processes (processes that are waiting for CPU scheduling, and not blocked / yielding). If the physical CPU is being allocated to processes on other virtual machines more often, then the load will go up, since more of your processes will queue up as runnable.
I hadn't considered this as a possibility. Thanks for the insight.
@perennate gave a good explanation. If your virtualmin install was idle, though, it might not be the case as loads over 20 on processes doing nothing is very uncommon, the load on the node must have been thousands.
Well, today, I checked my email account and saw two messages regarding load, again:
Time: Wed Mar 26 08:55:35 2014 -0400 1 Min Load Avg: 23.41 5 Min Load Avg: 15.28 15 Min Load Avg: 7.25 Running/Total Processes: 2/87
Time: Wed Mar 26 15:12:23 2014 -0400 1 Min Load Avg: 17.35 5 Min Load Avg: 8.40 15 Min Load Avg: 3.37 Running/Total Processes: 1/85
I won't beat up on this host because they have already gotten beaten up badly here. I had gotten this server as a cheap test bed for some ideas I had been kicking around but, honestly, I don't think it would be useful for me at this point. So, I canceled the service.
Oh, well.