Issues since upgrading from 10.3.1 to 11.0.9 (hostname and high CPU usage)

[NYCreate]

Hi all, hoping someone can help..

to start with I have very limited knowledge of linux or plesk so be gentle! ;-)

After upgrading I put IP through traceroute and it resolves to the wrong hostname.
Ive checked the hosts file, no error and even disabled DNS on the domain for which its showing as the hostname to see if this helped. Also checked hostname in server settings.

Any Ideas?..

Also as with a few other I see Im experiencing unusually high CPU loads (particularly high SQL cpu load) since installing nginx. (I presumed this was meant to help with things like this) I don't have a clue where to start and trying to resolve this one. Ive read another thread on the problem but it just baffles me..

Im hoping there are some technical peeps outthere that can help in 'not so technical' speak. I can follow most things however.

Thanks in advance

 

[abdi]

Change of Hostname

You can change easily the hostname by typing the command:

hostname NEW_NAME

or alternatively, login to plesk -> go to Tools & Settings -> Server Settings

 

[Nikolay.]

Don't just use the 'hostname' command, because:
1) the change will not be permanent on many of the systems;
2) Plesk won't know about the change if it was already configured before;
3) it won't help you anyway.

Do you have any real problem with the fact that reverse lookup may yield not the hostname of your machine? The way reverse DNS lookup works in the presence of multiple PTR records with same IP is that the end user (traceroute in this case) depending on it's logic will often receive only one name record. It will either be the "first" one or effectively a randomly chosen one. Do you have any problems with mail delivery to external servers?

High CPU load of any RDBMS is usually due to either ineffective (web) applications running on your server, high number of incoming requests, or (more rarely) misconfiguration. NGINX (at least in the current setup) will not help decrease or increase CPU load of any DBMS backend system.

 

[NYCreate]

You can change easily the hostname by typing the command:

hostname NEW_NAME

or alternatively, login to plesk -> go to Tools & Settings -> Server Settings

Hi, I did mention in the initial post I had checked the server settings and It is correct. thanks.

 

[NYCreate]

Don't just use the 'hostname' command, because:
Do you have any real problem with the fact that reverse lookup may yield not the hostname of your machine? The way reverse DNS lookup works in the presence of multiple PTR records with same IP is that the end user (traceroute in this case) depending on it's logic will often receive only one name record. It will either be the "first" one or effectively a randomly chosen one. Do you have any problems with mail delivery to external servers?

It seems a bit stupid a panel upgrade would actually affect something like this..
Ive disabled the DNS of the domain its using as the hostname but this didnt have any affect
The traceroute uses the same hostname everytime so I don't think its random. However, the domain it chose for the hostname is the last domain/subscription I added to plesk.

So far I can't see any issue with mail delivery, But I would like to correct the hostname issue.

Regards

 

[NelsonMuntz]

High CPU Usage

I upgraded my CloudLinux based Plesk system from Plesk 10.4.4 to 11.0.9 on Friday night and installed nginx immediately after the upgrade.

Like NYCreate, I am now seeing extremely high CPU usage, both from MySQL and Apache (system is consistently running at 60-70% CPU according to top). I also checked the virtual machine control panel supplied by my hosting provider and this very clearly shows a steady increase in CPU usage immediately after the upgrade, which only plateau'd when the system was nearly utilising all of the allocated virtual cores (8 of). I even went to the point of taking the main site on my server offline and confirmed there was no drop on utilisation.

This is clearly a result of the upgrade and not a poorly performing site (I do have a poorly performing site, which was part of the reason for installing nginx, but it was certainly not causing what I'm seeing now!)

Anybody else seen this yet and have a solution?

Thanks for any input!

 

[NelsonMuntz]

Incidentally, I've reviewed the items mentioned here, but none fit the bill. The server has also been rebooted with no improvement occurring.

 

[NelsonMuntz]

Another data point - I just stopped the reverse proxy configuration (as per last section here: http://download1.parallels.com/Ples...inistrator-guide/index.htm?fileName=70837.htm). The CPU load initially dropped noticeably (although not to previous levels), but is back up again unfortunately.

Typical top output:

top - 22:31:00 up 2:32, 1 user, load average: 6.61, 6.14, 6.11
Tasks: 227 total, 8 running, 219 sleeping, 0 stopped, 0 zombie
Cpu(s): 38.1%us, 46.6%sy, 0.0%ni, 15.0%id, 0.1%wa, 0.0%hi, 0.0%si, 0.1%st
Mem: 12521472k total, 5033772k used, 7487700k free, 341524k buffers
Swap: 1048568k total, 0k used, 1048568k free, 1794316k cached

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
6505 mysql 15 0 439m 47m 5172 S 117.8 0.4 132:53.56 mysqld
213494 apache 16 0 462m 89m 5360 R 39.4 0.7 1:01.26 httpd
219141 apache 16 0 462m 89m 5360 R 38.7 0.7 0:55.64 httpd
215989 apache 16 0 476m 91m 5368 R 36.7 0.7 0:54.38 httpd
213497 apache 17 0 462m 90m 5312 S 33.4 0.7 1:01.07 httpd
213631 apache 17 0 469m 90m 5380 S 29.8 0.7 1:05.09 httpd
229643 apache 16 0 462m 86m 5228 S 20.8 0.7 0:32.10 httpd
...

 

[NelsonMuntz]

Problem resolved itself

Bizarrely, this problem seemingly rectified itself after another day of high usage. Currently back at normal levels.

For what it's worth, I can't say I noticed a significant difference in CPU utilisation levels with nginx switched on or off after the above problem cleared itself up. However, we found that some users were unable to access our site homepage as they were being redirected to port 7080 for some reason. Haven't looked into what was causing that (certainly no explicit configuration done on the server) so have disabled nginx again for now anyway.

 

[abdi]

Current default nginx configuration does not help really much in the reduction of CPU or memory since almost all requests are passed to Apache.

However if you tune it further you can surely see a major difference.

 

[NelsonMuntz]

Thanks abdi. I'll investigate further.