Can you help me decipher this?

Still having on and off troubles with my hosts server speed. Their default answer is to blame slow sql queries then when you prove to them that there were only 3 slow queries in the last hour, then they go "wait a minute" and put me on hold for 10 - 20 minutes.

Then they tell me that I have too many processes running. So I ssh into the system and run top and get the following.

[email protected] [~]# top -u user
top - 12:11:26 up 92 days, 19:33,  4 users,  load average: 28.51, 46.37, 74.41
Tasks: 1554 total,   9 running, 1507 sleeping,   0 stopped,  38 zombie
Cpu(s): 72.7%us, 14.4%sy,  4.1%ni,  8.0%id,  0.5%wa,  0.0%hi,  0.2%si,  0.0%st
Mem:  33012580k total, 31901192k used,  1111388k free,   978424k buffers
Swap: 25164792k total,  1277252k used, 23887540k free, 21180840k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
30640 user  20   0  206m  18m 6764 S  5.6  0.1   0:00.17 php5
28971 user  20   0 13812 2264  828 R  1.6  0.0   0:00.52 top
10973 user  20   0 84484 1468  920 S  0.0  0.0   0:00.04 sshd

My questions are the following:
Obviously the tasks showing are me logged into the server but the task line shows 1554 tasks running. Are those MY tasks or the whole server?
The memory line: Is that MY memory usage or the whole server?

Looks to me like there are too many sleeping and zombie processes using up all the servers memory. Again are these processes MINE or the whole server?
:lamo:
JOhn ><>

Stuff4Toys:

the task line shows 1554 tasks running. Are those MY tasks or the whole server?

If a 'bare metal' server, then it'll be for the whole server. If a virtualised server it'll be for your VM.
Although it is impossible for me to say for sure if this is indicative of a problem (without more information), this is about 10 times more than I would normally expect (in either case)

Stuff4Toys:

The memory line: Is that MY memory usage or the whole server?

Probably the whole server/VM, but these figures are subject to manipulation in a VM environment.

Stuff4Toys:

Looks to me like there are too many sleeping and zombie processes using up all the servers memory.

Zombie processes don't use any memory, The are 'dead' pointers.
Sleeping processes are (in simple terms) processes that are not being accessed by the CPU at the instant the data is being read. Typically this will be the the total number of processes, minus the running, stopped and zombie processes

Stuff4Toys:

Again are these processes MINE or the whole server?
:lamo:
JOhn ><>

Based on the output data that you have provided, these figures would be for the whole server. If it were for the user named 'user' the total tasks would be '3' (the same number of processes listed in the second section of the TOP output).

Your load average is way off the scale for acceptable performance. The abnormally high number of zombies is probably a fallout of the system not being able to handle all of the running processes, and to solve the issue you need to somehow find out what all these processes are.

As stated, the number of processes is about 10 times greater than I would expect, BUT, if the server is under some kind of attack (ddos, spam), or, even if the server processing a maillist then these kinds of peaks are not unusual. Occasionally these peaks are the result of software that fails to close network or SQL sockets. Again, you will need to determine which program/app is causing this.

Try using the 'ps' command to see if that provides further clues.

Cheers
Rod

top - 12:11:26 up 92 days, 19:33,  4 users,  load average: 28.51, 46.37, 74.41
Tasks: 1554 total,   9 running, 1507 sleeping,   0 stopped,  38 zombie
Cpu(s): 72.7%us, 14.4%sy,  4.1%ni,  8.0%id,  0.5%wa,  0.0%hi,  0.2%si,  0.0%st
```The above 3 lines tell the story. This Server is overloaded.