Server Limit Error

Mick

Administrator
Staff member
Administrator
Joined
Jan 19, 1999
Messages
32,153
Reaction score
9,213
Sorry Guys,

A school boy error on my part I was looking for a file on our server with the words OPCACHE in it and for some mad reason I did this:

egrep -ir "opcache" /

I was supposed to be looking in the /etc folder!
you can imagine it searched the whole disk, and we have terabytes of data :(

which looked in every file on the system, loads on bash was 150+ and our server limit is 5
It was only because Rat sent me a message that I noticed ... I thought it was quiet last night :eek:

Sorry
Mick
 
Sorry Guys,

A school boy error on my part I was looking for a file on our server with the words OPCACHE in it and for some mad reason I did this:

egrep -ir "opcache" /

I was supposed to be looking in the /etc folder!
you can imagine it searched the whole disk, and we have terabytes of data :(

which looked in every file on the system, loads on bash was 150+ and our server limit is 5
It was only because Rat sent me a message that I noticed ... I thought it was quiet last night :eek:

Sorry
Mick

your bad boy 5 lashing for you
 
I think some additional tuition required :nopity:
 
I think some additional tuition required :nopity:


I FOUND THIS ONLINE:

I did a little and found this little GEM, and installed it on a every 10 minute cron...

Code:
You don't have permission to view the code content. Log in or register now.

This will email me when the server goes above load of 5

Cheers
Mick$trigger
 
@Mick

when you say a load above 5 what does that mean ?
 
Last edited by a moderator:
@Mick

when you say a load above 5 what does that mean ?

Generally you work out the server load based on your CPU (threads/cores)

IE on Centos you

grep pro /proc/cpuinfo -c

This will output your CPU codes/threads in our case it is 32 (which means we have 32 processors threads that can be used by a process).

Or the "top" command then press 1 will tell give you the following (this is ours just now).

Code:
You don't have permission to view the code content. Log in or register now.


Basically LOAD is what your processors can handle ours is 32 because we have 32 cores we can call on.
We also have 128GB RAM

However a realistic load for us is below 1, with 2000 online within the hour.

So you have to factor many things, how many people are online, how much ram is being used for each process - the ram is important because it will allow the process to be closed quicker - our average php call is about 10MB - 24MB so you times your max limit (2000) by the max RAM nedeed (24), so for 2000 users online at once we would need 48GB (64) just to keep the memory happy.

For me, apart from when the server is backing up, or cleaning the database (once a week).. I do not want to see it go over 5 However loads above 32 would be really worrying

Right now our server:
top - 14:06:10 up 74 days, 18:51, 1 user, load average: 0.43, 0.58, 0.67
 
Last edited:
Computers eh? It happens, I'm inspired to write about my recent 'interludes'...
 
Generally you work out the server load based on your CPU (threads/cores)

IE on Centos you

grep pro /proc/cpuinfo -c

This will output your CPU codes/threads in our case it is 32 (which means we have 32 processors threads that can be used by a process).

Or the "top" command then press 1 will tell give you the following (this is ours just now).

Code:
You don't have permission to view the code content. Log in or register now.


Basically LOAD is what your processors can handle ours is 32 because we have 32 cores we can call on.
We also have 128GB RAM

However a realistic load for us is below 1, with 2000 online within the hour.

So you have to factor many things, how many people are online, how much ram is being used for each process - the ram is important because it will allow the process to be closed quicker - our average php call is about 10MB - 24MB so you times your max limit (2000) by the max RAM nedeed (24), so for 2000 users online at once we would need 48GB (64) just to keep the memory happy.

For me, apart from when the server is backing up, or cleaning the database (once a week).. I do not want to see it go over 5 However loads above 32 would be really worrying

Right now our server:
top - 14:06:10 up 74 days, 18:51, 1 user, load average: 0.43, 0.58, 0.67


I think @Napster 's Brain just exploded. I know mines did lol.
 
Last edited by a moderator:
It means sfa to me also.

As long as someone knows what they're doing. :)

We shall see lol, I have updated all server software today...

I will be keeping an eye out

Mick
 
I upgraded the site to HTTP2 (which has been about since 2015, but now is supported by all major browsers).

HTTP2 is supposed to multi layer the site and pull all information at the same time.

How has the experience been since my last post on this ([FONT=&quot]25th February 2017, [/FONT][FONT=&quot]20:34), hopefully the site should be much quicker to navigate?

Let me know please guys, need some feedback...

Thanks you.

Mick




[/FONT]
 
@Mick I know it wouldn't of been of help in this instance as you didn't know there was a problem until Rat told you, but how about setting up a facebook page? Then if there are any problems with servers /downtime etc you can post on the facebook page to let us know that there is a problem and when the site may be back up etc.

I'm not saying posts all the time etc as I detest facebook lol,but as a tool solely for admin/mods to post re problems with here.
 
Last edited by a moderator:
Iv'e never really felt the site was slow,but deffo seems to be running smartish now
 
Back
Top