Virtual server loading slowly, not sure if Proxmox hardware needs updating?

[eiger3970]

Hello, I'm not sure if my Proxmox hardware/software configuration is slowing down my virtual server, or if the virtual server is at fault.

Any suggestions how to make Proxmox or the virtual server run quickly? (webpages are taking forever to load).
I'm worried that if I buy and/or allocate more RAM to the virtual server, that there may be some virus just eating up however much RAM is added.

I would also like some suggestions on setting up the email server.

 

[Tapio Lehtonen]

You have provided no information, so it is not possible to give any meaningfull answers.

 

[eiger3970]

Proxmox:

root@proxmox:~# df -h
Filesystem        Size  Used Avail Use% Mounted on
udev               10M     0   10M   0% /dev
tmpfs             3.2G  8.8M  3.2G   1% /run
rpool/ROOT/pve-1   94G  1.6G   92G   2% /
tmpfs             7.9G   40M  7.8G   1% /dev/shm
tmpfs             5.0M     0  5.0M   0% /run/lock
tmpfs             7.9G     0  7.9G   0% /sys/fs/cgroup
/dev/sde1         110G  969M  104G   1% /disc5storage
rpool              92G     0   92G   0% /rpool
rpool/ROOT         92G     0   92G   0% /rpool/ROOT
rpool/data         92G     0   92G   0% /rpool/data
/dev/fuse          30M   16K   30M   1% /etc/pve

https://photos.app.goo.gl/C8vJ4AQ9bzT9KbG98
https://photos.app.goo.gl/tbdZwW163mSiXj838
Webserver:
https://photos.app.goo.gl/Kfu5mVi6qtSpLA1A8
CMS:
Wordpress loads pages too slowly on website and on admin to edit pages.

 

[Tapio Lehtonen]

Seems web server is running CentOS 7, and has 2 GB memory. Not much to go by.

Type on the proxmox host command line this:

pveperf

Preferably when nothing is running.

 

[eiger3970]

Thanks, I think the issue is the webserver for the slow loading page.
I increased the webserver's RAM from 2 GiB to 4 GiB.
If more is needed, I'll upgrade, as I would like this running so I can move on to email servers for the website.

root@proxmox:~# pveperf
CPU BOGOMIPS:      24743.68
REGEX/SECOND:      2687833
HD SIZE:           93.07 GB (rpool/ROOT/pve-1)
FSYNCS/SECOND:     51.82
DNS EXT:           30.68 ms
DNS INT:           29.64 ms (com)
root@proxmox:~# df -h
Filesystem        Size  Used Avail Use% Mounted on
udev               10M     0   10M   0% /dev
tmpfs             3.2G  8.8M  3.2G   1% /run
rpool/ROOT/pve-1   94G  1.6G   92G   2% /
tmpfs             7.9G   43M  7.8G   1% /dev/shm
tmpfs             5.0M     0  5.0M   0% /run/lock
tmpfs             7.9G     0  7.9G   0% /sys/fs/cgroup
/dev/sde1         110G  969M  104G   1% /disc5storage
rpool              92G  128K   92G   1% /rpool
rpool/ROOT         92G  128K   92G   1% /rpool/ROOT
rpool/data         92G  128K   92G   1% /rpool/data
/dev/fuse          30M   16K   30M   1% /etc/pve

 

[Tapio Lehtonen]

If you run only that one Wordpress site on that virtual server / proxmos host, it should be enough according to pveperf. Put enough memory, and it should run just fine.
If you can get faster disk, that would make the server go faster. 50 IOPS is typical for one SATA disk system, so I assume that is what you have.
If you want to follow the resource usage on that CentOS host, install dstat and use it for example like this:

dstat --disk-tps --disk-util --dstat-cpu --mem --load --cpu -C total -s 10 360

 

[eiger3970]

Thanks for the handy tool. Still not sure why the WordPress website is running slow.
I suspect the Proxmox webserver is at fault (to handle a slow or heavy WordPress site).
I doubled the RAM.
I installed W3 Total Cache on WordPress to speed up the site.
I deactivated all plugins.
Still running slow?

Here's the results:

root@proxmox:~# dstat --disk-tps --disk-util --dstat-cpu --mem --load --cpu -C total -s 10 360
Terminal width too small, trimming output.
-dsk/total- sda--sdb--sdc--sdd--sde--zd0--zd16-zd32 --dstat-cpu--- ------memory-usage----- ---load-avg--- ----swap--->
reads writs|util:util:util:util:util:util:util:util|user syst tota| used  buff  cach  free| 1m   5m  15m | used  free>
   0    64 |4.30:4.30:4.29:4.32:0.00:   0:0.78:0.00|   0    0    0|5173M  304k  110M 10.5G|0.37 0.13 0.04|   0  8192M>
   1    79 |5.08:4.60:4.60:4.72:   0:   0:1.24:   0|   1    0    1|5174M  304k  110M 10.5G|0.55 0.17 0.05|   0  8192M>
   0    77 |4.44:4.12:4.12:4.12:   0:   0:   0:   0|   1    0    1|5174M  304k  110M 10.5G|0.62 0.20 0.06|   0  8192M>
   0   105 |4.40:4.12:4.16:4.28:   0:   0:0.20:   0|   0    0    0|5174M  304k  110M 10.5G|0.68 0.23 0.07|   0  8192M>
   0    64 |4.00:4.04:3.96:4.08:   0:   0:   0:   0|   1    0    1|5174M  304k  110M 10.5G|0.73 0.25 0.09|   0  8192M>

 

[eiger3970]

By the way, I don't used RAID due to server grade hardware. I'm not sure if Proxmox uses soft or hard caching. Could this be an issue with the slow website?

 

[Tapio Lehtonen]

Let us start from the beginning. What kind of host is proxmox installed on?

• Cpu
• ram
• Disk system
• Disk controller
• Is ZFS the file system? How is it set up?

Then the same for the virtual host.
What all stuff is running on proxmox and on the virtual host?
If you have other usefull information to assess the system performance, provide that also.

 

[eiger3970]

Proxmox is running: Webserver, discstorage (proxmox), local (proxmox) and local-zfs (proxmox).
CPU usage: 0.23% of 4 (CPU)s IO delay
Load average: 0.00,0.08,0.11
RAM usage: 32.60% (5.09 GiB of 15.62 GiB) KSM sharing
HD space(root): 1.79% (1.66 GiB of 93.03 GiB) SWAP usage
CPU: 4 x Intel(R) CPU E3-1220 v3 @ 3.10GHZ (1 Socket)
Kernel version: Linux 4.4.35-1-pve #1 SMP Fri Dec 9 11:09:55 CET 2016
PVE Manager Version: pve-manager/4.4-1/eb2d6f1e

Device        Type    Usage        Size              GPT    Vendor    S.M.A.R.T.    Wearout
/dev/sda     SSD     ZFS           111.79 GiB    Yes     ATA         PASSED       13%
/dev/sdb     SSD     ZFS           111.79 GiB    Yes     ATA         PASSED       0%
/dev/sdc     SSD     ZFS           111.79 GiB    Yes     ATA         PASSED       2%
/dev/sdd     SSD     ZFS           111.79 GiB    Yes     ATA         PASSED       0%
/dev/sde     SSD     mounted    111.79 GiB    No      ATA         PASSED       23%

Virtio generic disk controller.


Virtualhost:
Operating system: CentOS Linux 7.5.1804
Webmin version: 1.881
Usermin version: 1.741
Virtualmin version: 6.03
Theme version: Authentic Theme 19.09.2
Time on system: Wednesday, June 27, 2018 8:17 PM
Kernel and CPU: Linux 3.10.0-862.3.3.el7.x86_64 on x86_64
Processor information: Common KVM processor, 1 cores
System uptime: 3 days, 11 hours, 05 minutes
Running processes: 130
CPU load averages: 0.09 (1 min) 0.04 (5 mins) 0.05 (15 mins)
Real memory: 677.99 MB used / 3.61 GB total
Virtual memory: 0 bytes used / 1.95 GB total
Local disk space: 5.07 GB used / 42.89 GB free / 47.96 GB total
Virtualservers: 1 WordPress website 51.64 MB or 1024 MB and 1 pure HTML site 608 kB of 1024 MB.
DNS domains: 2.
Virtualwebsites: 2.
SSL websites: 0.
Mail domains: 0.
Databases: 3.
Mail/FTP users: 2.
Mail aliases: 0.
Apache Webserver 2.4.6 running.
BIND DNS Server 9.9 running.
SSH Server running.
MySQL Database Server 5.5.56-MariaDB running.
PHP version 5.4.16, 7.0.27.

 

[Tapio Lehtonen]

PHP folks claim PHP 7 is ten times faster than PHP 5. So there is a possibility to improve speed.
Is CentOS 7 an oldish version? Kernel and PHP seen a bit old.
Cpu in VM you have set to KVM. Set it to host, that is the fastest.
Other than that it looks like a very performant host. How are you determining Wordpress is slow? How fast do you expect it to be?
But disk system seems underperforming if you get only 50 iops with four ZFS disks plus SSD. How is it set up?

 

[Tapio Lehtonen]

I realized all disks are SSD. disks should definitely be faster than 50 iops. Was your measurement done while host was completely idle?

 

[eiger3970]

Thanks, I changed the CPU type from KVM to Host. Wordpress site and admin page still loading slowly.
Measurements were done with the Proxmox and the VM running.

 

[eiger3970]

So, the site is still slow. I suspect a DNS issue on Virtualmin, however Virtualmin says Virtualmin is fine.
Would Proxmox paid support be able to fix this?

 

[udo]

So, the site is still slow. I suspect a DNS issue on Virtualmin, however Virtualmin says Virtualmin is fine.
Would Proxmox paid support be able to fix this?

Hi,
I'm not sure that the proxmox support can fix your issue - for me it's looks for an IO-bottleneck.

you have 4 ssds for zfs and the first have an wearout of 13 %?! This smells like consumer-ssd?!

With enterprise ssd you get values like this:

pveperf
...
HD SIZE:           299.30 GB (rpool/ROOT/pve-1)
FSYNCS/SECOND:     4111.46
...

Udo