"Normal" load averages?

[heavyharmonies]

For the last week I've been seeing what I perceive to be load averages higher than usual on my VPS. Normally the load average is below 1.0, occasionally spiking into the 2+ range. For the last week the load average has been staying in the 1-3 range on a regular basis, spiking over 5.

There has not been an associated increase in traffic.

When I bring this to the attention of support, I'm repeatedly being told not to worry about it and loads of 3+ are normal.

I disbelieve. This has NOT been my experience, and I'm concerned as it is translating into sluggish performance at the browser end.

I asked if the node I'm on (UL10) has seen greater usage as of late, but did not receive an answer.

[jason]

Have you recently done anything that might effect the average (sowftware updates, tweaked settings, etc.)? It could be that an automatic update changed something in the config of one of your daemon processes and that's causing a resource spike.

Just grasping at straws, though. It is difficult to tell what could be going on without having access to the server and knowing its history.

--Jason

[heavyharmonies]

No changes have been made by me in well over a year. If anything has been tweaked by Jag, I'm unaware of it.

I'm an "if it ain't broke, don't fix it" kinda guy. I don't tweak and adjust things on my VPS. About the only thing I do is occasionally install a perl module.

That's why I get nervous when I see performance changes... I know it's nothing I've changed...

[Daniel_DBS]

I noticed higher load averages than normal on my VPS node as well... Normal is around 0.5 or less... The 15 minute average for a day or two was a constant 1.5+

I thought this was a bit strange so I let Support know and they did something and admitted it was a bit higher than normal, and now loads are down to almost zero!

I am not sure what they did but the highest it has been since i let them know a couple days ago was 0.2 and at times it is at 0.00!

I love it!

[heavyharmonies]

*harumph*

Lucky you. Load on my VPS is 4.45 as I type this. It's been above 2.0 all morning.

Just out of curiousity, what kind of traffic do the domains on your VPS receive?

[Daniel_DBS]

Heh? Not that much traffic to be honest!

I think my busiest personal site receives about 200 hits per day. Each month on my VPS I HARDLY use a fraction of my allotted 750GB of monthly bandwidth... Last month I think I used less than 20GB

My goal is not to have extremely busy sites... It's mostly the learning experience... A couple months ago you could ask me what php was and I would say HUH?

[wuurp]

In my experience (on UL9), if I see a 15 minute load average over 1.0 I start looking to see what's wrong. This includes the occasional day when I hit over 50GB traffic. It's often associated with high iowait values.

I have noticed that in the last week, vzpp reports a number of failed attempts to migrate my vps. In a support ticket it was suggested that since UL9 is an old DEHE node, that someone is trying to migrate me to a new node. Maybe you're already on a faster, better server.

[heavyharmonies]

10:29:07 up 2 days, 18:52, 2 users, load average: 6.68, 5.23, 5.16

I think I see what's killing the VPS... my weekly backups are running... now why they're configured to run during normal business hours instead of say 1:00am, I have no friggin' idea (Jag configured the backups).

As soon as the backup is done, I'll be filing a ticket to get that changed...

[Daniel_DBS]

My backup routine runs about 2AM est.

Even when my backups are running the server hardly even breaks a sweat!

[heavyharmonies]

Stupid question: Shouldn't 'ps x' tell me how long each of the processes has been running? The time reads 0:00 for every single process, including the backup process.

I want to see just how long the weekly backup has been running...

[Daniel_DBS]

I thought that was 'ps -A'

[heavyharmonies]

That shows 00:00:00 for all processes...

[Daniel_DBS]

Hehehe:

10:47:35 up 1 day, 15:41, 1 user, load average: 0.02, 0.45, 0.53

[Daniel_DBS]

That shows 00:00:00 for all processes...

Hmmmm! It doesn't on mine! It actually shows the time a process has been running!

[wuurp]

In my experience (on UL9), if I see a 15 minute load average over 1.0 I start looking to see what's wrong. This includes the occasional day when I hit over 50GB traffic. It's often associated with high iowait values.

I have noticed that in the last week, vzpp reports a number of failed attempts to migrate my vps. In a support ticket it was suggested that since UL9 is an old DEHE node, that someone is trying to migrate me to a new node. Maybe you're already on a faster, better server.

Well, my vps was successfully migrated to UL12. Now my load average is 3 or 4 times what it typically was, and my site is sluggish, with only a couple visitors. Maybe there were only a couple people left on UL9 and I got spoiled. I wish they would have left me there.