I have an account (shared hosting) at venturesonline to host a (little) vBulletin forum.

Each time I run the 'top' command it shows something like this :

7:15am up 14 days, 14:12, 4 users, load average: 6.36, 4.48, 3.62
403 processes: 389 sleeping, 7 running, 6 zombie, 1 stopped
CPU states: 31.4% user, 63.6% system, 0.0% nice, 4.8% idle
Mem: 2009636K av, 1976044K used, 33592K free, 3598900K shrd, 422816K buff
Swap: 530104K av, 6488K used, 523616K free 343800K cached

(There are 2 cpus on this server)

Is the load high ? I thought that venturesonline was a top-quality host...

Originally posted by ABW
CPU states: 31.4% user, 63.6% system, 0.0% nice, 4.8% idle

That's a pretty busy CPU. Is the load always that high? Maybe they were running stats or something when you checked it. I think someone once mentioned that the general rule of thumb is to try to keep the load averages below 5.00.

Yeah always... sometimes it goes down to 2 - 2.5 but never below :eek:

Just done another 'top' right now :


8:17am up 14 days, 15:14, 4 users, load average: 6.09, 4.75, 4.32
488 processes: 475 sleeping, 4 running, 8 zombie, 1 stopped
CPU states: 35.0% user, 61.7% system, 0.0% nice, 3.2% idle
Mem: 2009636K av, 1960740K used, 48896K free, 783240K shrd, 318904K buff
Swap: 530104K av, 6484K used, 523620K free 115876K cached

Originally posted by ABW
Yeah always... sometimes it goes down to 2 - 2.5 but never below :eek:

Just done another 'top' right now :


8:17am up 14 days, 15:14, 4 users, load average: 6.09, 4.75, 4.32
488 processes: 475 sleeping, 4 running, 8 zombie, 1 stopped
CPU states: 35.0% user, 61.7% system, 0.0% nice, 3.2% idle
Mem: 2009636K av, 1960740K used, 48896K free, 783240K shrd, 318904K buff
Swap: 530104K av, 6484K used, 523620K free 115876K cached

This is just a guess, but being that you are on a shared account, I would guess that the server has too many accounts on it.

The server has plenty of CPU Power (2) and RAM (2 GB!) to handle quite a load of processing. So I doubt it is the server. Chances are it is just always running user CGI processes since there are so many accounts on there.

Perhaps you should ask to be moved to another server?

Take care,

Brian

Being new at this... can someone tell me how to do that 'top' command on a shared hosted account? Do I need root access?
Thanks for your detailed responses
Antonio

Global: you'd need either telnet or SSH access.

Try start -> run -> telnet -> connect -> remote system

Enter your domain as the hostname

Then it should ask you for your username & password if you have telnet access. If it lets you in, simply type top

You don't need root access.

login with telnet/SSH and type: top

that's about all the details i can give :)

And another one...

9:31am up 14 days, 16:28, 7 users, load average: 8.18, 6.52, 5.17
542 processes: 526 sleeping, 7 running, 8 zombie, 1 stopped
CPU states: 29.5% user, 48.3% system, 0.0% nice, 22.1% idle
Mem: 2009636K av, 1986196K used, 23440K free, 793176K shrd, 68072K buff
Swap: 530104K av, 6484K used, 523620K free 70992K cached


Perhaps you should ask to be moved to another server?


I'm going to ask them... or maybe I will leave VO completely and ask for money back...

I don't recommend them personally

I think they have a serious problem here...
It's so overloaded I can't do a 'top'
I tried 'uptime' and look at the result :

9:42am up 14 days, 16:38, 7 users, load average: 111.02, 47.27, 21.10 :eek: :eek:

Somebody else, icqed me and showed me that their hard drives were full on one of their servers, i wonder why they dont monitor their resources??

I think they might overload their machine..
a while ago iwas with some other host, I had the same problem. They put well over 600+ accounts per server. So I cancelled my account a few days later.

Well.... a lot of people recommend VO.. :confused:

I'm really disappointed and I think that people who recommend them don't have a shared account with them (maybe they have dedicate servers...)

Originally posted by ABW
Well.... a lot of people recommend VO.. :confused:

I'm really disappointed and I think that people who recommend them don't have a shared account with them (maybe they have dedicate servers...)

I think the people that recommend them is because of their cheap reseller prices with tons of bandwidth. Never heard anyone say how fast their servers are.

Originally posted by ABW
I think they have a serious problem here...
It's so overloaded I can't do a 'top'
I tried 'uptime' and look at the result :

9:42am up 14 days, 16:38, 7 users, load average: 111.02, 47.27, 21.10 :eek: :eek:

Damn!!! 111 waiting, run queue processes? That has to be a record.

Now I don't think there's any doubt THAT server is overloaded. Time to get on another server or look elsewhere!

Good luck,

Brian

ABW

I suggest u email them with ur concern. I think they probably will put u on a less busy server!

:eek: what in da world is that server running??

Now the server has come back to 5-6 of load.

Here's a screenshot where you can see the record is :eek: 213.73 :eek:

I couldn't do the 'top' to see what was causing this...

maybe you should try posting in VO's forums about the high load on your server?

You will get your problem fix allot quicker if you contact the host. Although this is a good place to ask for opinions, your problem wouldn't be fix by us. Give them a chance to fix your problem before leaving. Good luck and I hope you get this resolved.

Originally posted by cahostnet
You will get your problem fix allot quicker if you contact the host. Although this is a good place to ask for opinions, your problem wouldn't be fix by us. Give them a chance to fix your problem before leaving. Good luck and I hope you get this resolved.

I already contacted them about this issue.

I posted this thread here to share this "record load" of 213 :)

What are they doing about it?

I got this off 1 of our accts there...

12:02pm up 9 days, 13:47, 2 users, load average: 1.30, 1.50, 1.53
247 processes: 239 sleeping, 6 running, 2 zombie, 0 stopped
CPU states: 35.1% user, 22.7% system, 0.0% nice, 42.0% idle
Mem: 2074332K av, 1301048K used, 773284K free, 2103952K shrd, 629172K buff
Swap: 265032K av, 1852K used, 263180K free 230740K cached

which server are you on?

A load of 5 with 2 CPUs isn't really that bad. Though I'm not sure why you think it has two CPUs. On the multiprocessor machines I've used top will show "CPU states" for each CPU independantly.

Now a load of 111 or 213 is certainly overloaded. I'd guess they have some with a cgi that's spinning and consuming all the resources. They should be able to track that down and turn off the offender. The reality of shared hosting is you can alway be affected by the other accounts on the machine. Even if there are only two accounts on there, the other guy could bring the machine to its knees.

A load of 5 on a dual isnt terrible but it probably shouldnt be there consistantly... thats really where they should be peaking at. The server theyre referring to most likely doesnt have a load of 100+ regularly, it is probably a freak thing happening.

-Brendan

Originally posted by Kaith Sutai-Rustaz
which server are you on?

I don't know on which server I am on, but we are not on the same (yours has been 'up' 9 days and the one I'm on has been 'up' for 14 days)

Current 'top'

1:33pm up 14 days, 20:29, 7 users, load average: 5.57, 4.54, 5.02
508 processes: 492 sleeping, 9 running, 6 zombie, 1 stopped
CPU states: 34.0% user, 32.7% system, 0.0% nice, 33.1% idle
Mem: 2009636K av, 1944648K used, 64988K free, 181160K shrd, 232880K buff
Swap: 530104K av, 23388K used, 506716K free 124480K cached


It seems to be normal...

Have you already mailed them????
I know that they put about 200-250 accounts on a big server like this, which should do fine. Maybe a user does massive CGI and abuses the server...

The load could be driven up by the large % of cpu time the system is using. I'd check to see what device is using so much cpu time. In most cases its the disk drive. You can watch and see what is devices are inturrupting the system with something like this:

watch -n 1 cat /proc/interrupts

Hi everyone,

I work for VO now!!

Ok, let's tackle the problem we had with Gamble today. We had several cases of cgi abuse on that server in the past week and are working hard to seek them out and suspend the offending accounts. You're right, a load of 113 is definitely too high, and this is definitely not normal.

If you are not aware, we use cpanel/whm on our servers, which resellers use to create new accounts. New accounts are created daily (if not hourly) on gamble, and while we are in the process of setting up a couple new servers (for shared/reseller hosting), gamble is taking the majority of new accounts, which also means the majority of new offenders.

Average load for gamble is between 2 and 4, which is definitely acceptable for dual CPU system with 2GB of ram. However, we agree that anything higher is not acceptable, which is why the new servers are going up and we will be offering current resellers the opportunity to move to a new box.

Some people may ask, 'well why did you wait to set up new boxes'. Well, we just finished our move from VDI and are finally settling in. We didn't want to setup a new box at VDI which would just have to be moved a couple of weeks later.

This along with seeking out Terms and Conditions abusers should keep Gamble's load acceptable.

Here's a snapshot of gamble's load right now:

2:22pm up 14 days, 21:18, 5 users, load average: 3.15, 3.38, 4.46
474 processes: 464 sleeping, 2 running, 7 zombie, 1 stopped
CPU states: 21.3% user, 24.4% system, 0.0% nice, 54.2% idle
Mem: 2009636K av, 1325728K used, 683908K free, 3275936K shrd, 157140K buff
Swap: 530104K av, 23360K used, 506744K free 122472K cached

As you can see, it's doing fine. There isn't much we can do with cgi abuse, other than seek them out and terminate their accounts. The problem is that, the damage has already been done at that point, hence the reason you made this post.

We sincerely apologize for any inconvenience this has caused you, and can assure you that we do not have a policy of overloading servers. A policy like this would only cause us more headaches and problems, something any decent host works to avoid.

Originally posted by ABW
Now the server has come back to 5-6 of load.

Here's a screenshot where you can see the record is :eek: 213.73 :eek:

I couldn't do the 'top' to see what was causing this...


what! :smash:

You might want to try running /scripts/apachelimits as it will put a ulimit on how much cpu and memory a process running from the webserver can use and keep your server from spiraling down to a cold death.


Originally posted by tabernack
Hi everyone,

I work for VO now!!

Ok, let's tackle the problem we had with Gamble today. We had several cases of cgi abuse on that server in the past week and are working hard to seek them out and suspend the offending accounts. You're right, a load of 113 is definitely too high, and this is definitely not normal.

If you are not aware, we use cpanel/whm on our servers, which resellers use to create new accounts. New accounts are created daily (if not hourly) on gamble, and while we are in the process of setting up a couple new servers (for shared/reseller hosting), gamble is taking the majority of new accounts, which also means the majority of new offenders.

Average load for gamble is between 2 and 4, which is definitely acceptable for dual CPU system with 2GB of ram. However, we agree that anything higher is not acceptable, which is why the new servers are going up and we will be offering current resellers the opportunity to move to a new box.

Some people may ask, 'well why did you wait to set up new boxes'. Well, we just finished our move from VDI and are finally settling in. We didn't want to setup a new box at VDI which would just have to be moved a couple of weeks later.

This along with seeking out Terms and Conditions abusers should keep Gamble's load acceptable.

Here's a snapshot of gamble's load right now:

2:22pm up 14 days, 21:18, 5 users, load average: 3.15, 3.38, 4.46
474 processes: 464 sleeping, 2 running, 7 zombie, 1 stopped
CPU states: 21.3% user, 24.4% system, 0.0% nice, 54.2% idle
Mem: 2009636K av, 1325728K used, 683908K free, 3275936K shrd, 157140K buff
Swap: 530104K av, 23360K used, 506744K free 122472K cached

As you can see, it's doing fine. There isn't much we can do with cgi abuse, other than seek them out and terminate their accounts. The problem is that, the damage has already been done at that point, hence the reason you made this post.

We sincerely apologize for any inconvenience this has caused you, and can assure you that we do not have a policy of overloading servers. A policy like this would only cause us more headaches and problems, something any decent host works to avoid.

Originally posted by tabernack
I work for VO now!!

OT: since when? :)

Originally posted by Walter


OT: since when? :)

This is my first week fulltime, but I've been doing what I can while I finished up my cubicle job :D

Originally posted by bdraco
[B]You might want to try running /scripts/apachelimits as it will put a ulimit on how much cpu and memory a process running from the webserver can use and keep your server from spiraling down to a cold death.
/B]

Great! Thanks Nick.

Well I can guarantee that box running to 100+ load was definately a fluke, as for it being over loaded, that it did happen slightly, we had an issue with one of our move servers since we were coming out of VDI which removed the replacement for that server and pushed it down the line, but the load should decrease shortly , we have a large reseller moving off of that server to his own and after today no new accounts will be going on it, the move has just finished so we are working to get everything straightened back out and into a smooth rythm.

Originally posted by bdraco
You might want to try running /scripts/apachelimits as it will put a ulimit on how much cpu and memory a process running from the webserver can use and keep your server from spiraling down to a cold death.




That's definitely better than nothing, but to limit all of Apache with the same user, will either put too much limits and result in errors and delays, or it will only allow Apache to still consume too much resources. Still, not a bad idea, better than nothing, but setting reasonable limits on each of the accounts would be more effective, not effect everyone, other than the abusive site/user/script(s) and not leave you with the choice of either having Apache slow or erroring, or having the system crash. Of course, by doing that... well, nevermind, I don't want to have people misunderstand me and start a problem..........

I have been watching this server very close today I also have suspended several accounts that was using a lot of resources. Right now its running at High but normal pace we will keep watching it thru out the night, we will also be moving accounts to one of our new servers in the morning so it will only get better from here.

Thanks Nick and Tim for the help :)

And to the rest we are working on this trust me.

Originally posted by ABW
I don't know on which server I am on, but we are not on the same (yours has been 'up' 9 days and the one I'm on has been 'up' for 14 days)

Current 'top'

1:33pm up 14 days, 20:29, 7 users, load average: 5.57, 4.54, 5.02
508 processes: 492 sleeping, 9 running, 6 zombie, 1 stopped
CPU states: 34.0% user, 32.7% system, 0.0% nice, 33.1% idle
Mem: 2009636K av, 1944648K used, 64988K free, 181160K shrd, 232880K buff
Swap: 530104K av, 23388K used, 506716K free 124480K cached


It seems to be normal...

None of the Top paste's posted here display a 2 cpu server though. This is what TOP shows when you have 2 cpu's, here is a top paste from one of our dual cpu servers:

6:56pm up 19 days, 18:52, 3 users, load average: 0.22, 0.25, 0.19
166 processes: 164 sleeping, 1 running, 1 zombie, 0 stopped
CPU0 states: 1.3% user, 3.2% system, 0.0% nice, 94.3% idle
CPU1 states: 9.1% user, 3.1% system, 0.0% nice, 87.1% idle
Mem: 2096664K av, 1291944K used, 804720K free, 0K shrd, 686476K buff
Swap: 136512K av, 0K used, 136512K free 180500K cached

Note the second line CPU line to indicate the load of each cpu.
Now I could be wrong but I have not ever seen TOP not read two CPU lines when you run dual cpu's. If I am wrong and its possible to make two appear as one then I apologize.

Not all TOP's will show that, 7.1 systems do but apparently the 6.2's do not.

Actually that is from a 6.2 running cpanel and does show both.

Originally posted by ABW


I already contacted them about this issue.

I posted this thread here to share this "record load" of 213 :)

It sounds like TOP is not reporting correctly. A load that high would surely shut off all services and not have the cycles left to even run TOP.

Originally posted by Jag

Note the second line CPU line to indicate the load of each cpu.
Now I could be wrong but I have not ever seen TOP not read two CPU lines when you run dual cpu's. If I am wrong and its possible to make two appear as one then I apologize.

It depends upon what version of top you're using. The BSD versions only show one line for all CPUs combined; older GNU versions follow this same behaviour while more recent GNU versions have a commandline option (-C I believe?) to control that.

Bahhhh

I rember back when WHT was on the "Wizard" server, one night a rogue script caused it to run for a couple of hours at 90% .. nobody noticed it .. not even the guys here at WHT :)

Never had any trouble or downtime, not even in the move to new NOC. Here is top from telnet for me:

6:51pm up 11 days, 20:37, 1 user, load average: 0.80, 1.16, 1.13
222 processes: 219 sleeping, 1 running, 2 zombie, 0 stopped
CPU states: 35.8% user, 12.8% system, 0.0% nice, 51.2% idle
Mem: 2074332K av, 1392316K used, 682016K free, 1901088K shrd, 446516K buff
Swap: 265032K av, 1472K used, 263560K free 306096K cache

Let me be the first to say, I have never had a good host so far. NEVER. I have watched the output from top so far and this is more or less an average.

Granted, that is only the server I am on. But its the best I've had so far, period.

:)

We never let our servers go above 1.5 on average.

Actually most of our servers are never above 1.0..

We always monitor our servers...

:-)

<<MOD NOTE:>>
And what does this have to do with the thread? Signature removed.
<</MOD NOTE>>

12:37pm up 21 days, 19:33, 6 users, load average: 11.34, 8.81, 7.46
588 processes: 571 sleeping, 8 running, 8 zombie, 1 stopped
CPU states: 57.6% user, 34.8% system, 3.8% nice, 3.5% idle
Mem: 2009636K av, 1986772K used, 22864K free, 1882868K shrd, 293024K buff
Swap: 530104K av, 14184K used, 515920K free 275468K cached


Again... and the load is like this every days... I really don't recommend anyone to go with venturesonline.

Well, if cpanel actually had a way to set the exact time for the logs to run there wouldn't be a problem, but they lovelingly like to run in the middle of the day which is the cause of this load, I am working to rectify it, and as explained earlier clients will be coming off of this machine, but your more than welcome to your own opinions.

you could maybe change the time on the server, maybe to uk time. so the logs wont run at the us peak time.

Yes but can you imagine the hundreds of bulletin boards that i would really throw for a loop if i did that

i'm going to attempt to use cpanel's builtin script to spark the logs a going at about 2 in the morning.... but if its going to stop them from running whenever they please i don't know.......

Nick knows i yelled @ him for not giving me a way to control the time that the logs started :-/

Originally posted by DanielP
Well, if cpanel actually had a way to set the exact time for the logs to run there wouldn't be a problem

I wonder what programs controlled by CPanel that actually runs during peak times? I have a cron job recording the CPU Load of my server every single minute and the only time CPU Load went up above our typical normal CPU Load was around 4:00 am (Definitely not a peak time). And I think during this period, CPanel runs webalizer and hence the CPU Load always goes up a bit ...

Cheers,
:beer:

If i'm correct, cpanel's logs run off of the /home/username/tmp/lastrun's time stamp instead of a conventional cron or time file so somehow the timestamp of those files has gotten tob e during the day on that one server but i should be able to rectify that tonight :)

No comment...

6:22pm up 22 days, 1:18, 8 users, load average: 82.20, 33.11, 14.63

6:23pm up 22 days, 1:19, 9 users, load average: 94.98, 42.32, 18.49

I dont think its the stats/log stuff doing that... it never raises the load on our servers above 5 or so even when it runs during peak usage times.

-Brendan

So is this good????
8:56pm up 7 days, 10:50, 4 users, load average: 0.14, 0.31, 0.33
133 processes: 128 sleeping, 2 running, 3 zombie, 0 stopped
CPU states: 17.0% user, 16.6% system, 9.8% nice, 56.3% idle
Mem: 516016K av, 494364K used, 21652K free, 423696K shrd, 97872K buff
Swap: 1028120K av, 168K used, 1027952K free 145080K cached
thanks in advance!

Well, ABW, had you put in a page to alert me rather than posting it here I might have found the source of the issue, but currently its purring along fairly nicley, but I can only really offer you one alternative, and that would be to move you to our new server which went up last week, aside from that all I can say is that accounts will be coming off that server but unfortinuately its not as simple as snapping my fingers and their off.

Originally posted by DanielP
unfortinuately its not as simple as snapping my fingers and their off.

I wish it was that easy. I wonder what kind of $/hour I could get if I had that gift :D

Originally posted by tabernack
I wish it was that easy. I wonder what kind of $/hour I could get if I had that gift :D
A whole lot!! I wish I could snap my fingers and hosts would disappeare coughcough*****coughTACIDcoughcough hehehe :D

If I remember correctly.....

The LoadAverage of 1.0 == to one processor.
so 1.0 == 100% .50 = 50%

So if you have 2 processors , 2.0 == 100%

so the load was about 6.0 , so that's 6.0/2.0 = 3.0 == 300% , it's running over 300% CPU LOAD.

YES, I THINK the server is having some problems.

I don't think they have any admin watching them processes. bad bad bad.


Quoc Le

And the point of your comment is?

We are quite well aware that there are a few too many accounts on that system but what has been said time and time again in this continuing loop of a thread is that accounts will be coming off to reduce the load but things take time

Originally posted by Quoc
If I remember correctly.....

The LoadAverage of 1.0 == to one processor.
so 1.0 == 100% .50 = 50%

So if you have 2 processors , 2.0 == 100%

so the load was about 6.0 , so that's 6.0/2.0 = 3.0 == 300% , it's running over 300% CPU LOAD.

YES, I THINK the server is having some problems.

I don't think they have any admin watching them processes. bad bad bad.


Quoc Le

This really is not exactly how it works. A dual server with a decent amount of ram can function with a load of six quite well. Above that it gets hairy.

-Brendan

As Daniel said If you would spend your time putting in a help desk ticket or paging us instead of posting your problems here this would get resolved a lot sooner, our Admins do not work at WHT.

If you had put in a ticket then you would know there is and open offer to move anybody off that server to one of our new servers that wants to.

If this is not a good solution to you and you wish to cancle then I will be happy to give you a full refund for this month.

If none of the above help you then its clear your just here to flame VO and nothing more.

Also to to clear a few things up we do have 5 full time and 1 part time admins that work on servers everyday they are not logged in watching every servers load at every moment but when a problem comes up they do check it out ASAP if reported.

What's up with you people!? Do you really need to post same things over and over again? You are aware of the problem, they are aware of the problem! There is no need to point it out every 2 minutes. They are offering a solution to the problem! Instead of wasting your time whining about it here (over and over), email them and have your account moved to the new server if the server load really bothers you!

As I have a few accounts with VenturesOnline I have to state that their level of service is very good.
I am not on the same server as you but high load happened to my server before and they did take care about.

Originally posted by Dogma
So is this good????
quote:
--------------------------------------------------------------------------------
8:56pm up 7 days, 10:50, 4 users, load average: 0.14, 0.31, 0.33
133 processes: 128 sleeping, 2 running, 3 zombie, 0 stopped
CPU states: 17.0% user, 16.6% system, 9.8% nice, 56.3% idle
Mem: 516016K av, 494364K used, 21652K free, 423696K shrd, 97872K buff
Swap: 1028120K av, 168K used, 1027952K free 145080K cached
--------------------------------------------------------------------------------

thanks in advance!

Actually Dogma, that amount of usage of your swap space indicates you may need more ram on that server, otherwise it looks good. Hope this helps.

Originally posted by Jag

Originally posted by Dogma
So is this good????
quote:
--------------------------------------------------------------------------------
8:56pm up 7 days, 10:50, 4 users, load average: 0.14, 0.31, 0.33
133 processes: 128 sleeping, 2 running, 3 zombie, 0 stopped
CPU states: 17.0% user, 16.6% system, 9.8% nice, 56.3% idle
Mem: 516016K av, 494364K used, 21652K free, 423696K shrd, 97872K buff
Swap: 1028120K av, 168K used, 1027952K free 145080K cached
--------------------------------------------------------------------------------

thanks in advance!
Actually Dogma, that amount of usage of your swap space indicates you may need more ram on that server, otherwise it looks good. Hope this helps.

Actually the swap usage is very good. There is only 168K used, so there is plenty of RAM. There is just 1 GB of swap space available. That is probably overkill, but that is how it appears to be setup.

Take care,

Brian

usage of swap space if the first key to let you know you need to watch the memory. If memory is more than sufficient the swap will rarely ever get used.

Thanks Jag and Bteeter!! Just wondering 'bout it...... :)