High CPU Useage

E

Eric535

Member
So i have 3 linodes with MW installed on each of them. 2 of them have the same cpu and memory and the other has a big CPU and memory. Whats weird is that the smaller linodes email faster and the CPU doesnt go up as high as the bigger linode. Below are the details of each linode. Keep in mind that the bigger one emails slower and the cpu gets up to 90% when I send a campaign. The only thing I did different was with the bigger linode, I setup MW using the bash command explained here https://kb.mailwizz.com/articles/install-mailwizz-using-just-one-bash-command/. The other two I used serverpilot.


Server 1 (Linode $5)
1 GB
1CPU
Sends 30k email in 1hr (500 per minute)
Using Sparkpost
Memory goes to about 49% while sending a campaign
CPU Stays around 11% while sending a campaign

Server 2 (Linode $5)
1 GB
1CPU
Sends 30k email in 1hr (500 per minute)
Using Sparkpost
Memory goes to about 49% while sending a campaign
CPU Stays around 11% while sending a campaign

Server 3 (Linode $10)
2 GB
1CPU
Sends 30500k email in 1.5hr (339 per minute)
Using Amazon SES
Memory goes to about 36% while sending a campaign
CPU Stays around 85% while sending a campaign
 
Can be dozen reasons, including the type of emails you're sending, if you embed images in them, if you dkim sign the emails from mailwizz, etc etc, so is hard to say the real reason.

All those linoes are super mega tiny, to me the performance seems acceptable.
 
Eric535 said:
So i have 3 linodes with MW installed on each of them. 2 of them have the same cpu and memory and the other has a big CPU and memory. Whats weird is that the smaller linodes email faster and the CPU doesnt go up as high as the bigger linode. Below are the details of each linode. Keep in mind that the bigger one emails slower and the cpu gets up to 90% when I send a campaign. The only thing I did different was with the bigger linode, I setup MW using the bash command explained here https://kb.mailwizz.com/articles/install-mailwizz-using-just-one-bash-command/. The other two I used serverpilot.


Server 1 (Linode $5)
1 GB
1CPU
Sends 30k email in 1hr (500 per minute)
Using Sparkpost
Memory goes to about 49% while sending a campaign
CPU Stays around 11% while sending a campaign

Server 2 (Linode $5)
1 GB
1CPU
Sends 30k email in 1hr (500 per minute)
Using Sparkpost
Memory goes to about 49% while sending a campaign
CPU Stays around 11% while sending a campaign

Server 3 (Linode $10)
2 GB
1CPU
Sends 30500k email in 1.5hr (339 per minute)
Using Amazon SES
Memory goes to about 36% while sending a campaign
CPU Stays around 85% while sending a campaign
Click to expand...
For more valid comparative data, check also database size, number of segments used in mailed list, cpu specs, and compare spark to spark or ses to ses ;)
 
twisted1919 said:
Can be dozen reasons, including the type of emails you're sending, if you embed images in them, if you dkim sign the emails from mailwizz, etc etc, so is hard to say the real reason.

All those linoes are super mega tiny, to me the performance seems acceptable.
Click to expand...

I sent the same type of emails.
 
twisted1919 said:
@Eric535 - You could try this: https://kb.mailwizz.com/articles/mysql-mariadb-optimisations/
And also maybe increase the size of those instances...
Click to expand...
@twisted1919 So if my /etc/my.cnf is empty I need to add those lines mentioned in the link? Right now I have a linode 2048. Is 2gbs of ram and 1core and 1 cpu good enough? Every time I send emails, my cpu cruises around 89%-99%. I have two other servers that are linode 1024's and they dont have this issue. The only differences I can think of, is the one with high CPU (linode 2048) i use UTM links in my campaigns and Amazon SES. The other ( linode 1024) i dont use UTM and I use sparkpost. I can get of 40k emails in a little over an hour. My linode 2048 is much slower in sending emails.
 
Eric535 said:
So if my /etc/my.cnf is empty I need to add those lines mentioned in the link?
Click to expand...
If the file is empty, your settings must go under a mysqld block, like:
Code: 
[mysqld]
max_connections...
etc
etc
Eric535 said:
Every time I send emails, my cpu cruises around 89%-99%. I have two other servers that are linode 1024's and they dont have this issue. The only differences I can think of, is the one with high CPU (linode 2048) i use UTM links in my campaigns and Amazon SES.
Click to expand...
Embedding images or adding attachments might also cause high cpu load.
 
twisted1919 said:
If the file is empty, your settings must go under a mysqld block, like:
Code: 
[mysqld]
max_connections...
etc
etc

Embedding images or adding attachments might also cause high cpu load.
Click to expand...


so in that same /etc/my.cnf file I would have some lines that go like

[mysqld]
settings1
settings2
setting3

Does it have a closing bracket?

The emails are the same between both servers No attachments or images in the actual email. Images are hosted. Do you think im going to have to upgrade to Linode 4096?
 
Is the file really empty? Maybe the MySQL config file is stored elsewhere?

I would start by doing this:

# find / - depth -print I grep my.cnf

---

@Fabyc , I just purchased a VPS with 6 cores, 8GB RAM (guaranteed, burst up to 12 GB), 200 GB SSD and free traffic. A measly €13/month.
OpenVZ though...
 
Last edited:
Fabyc said:
Hi @OptiBiz1

What provider is it?
Does it have 24/7 support?

Thanks
Click to expand...


Server4You, but if you are unlucky with what support technician you are in contact with, they suck *ss. I gotta tell you that.
I was their customer on dedicated servers for six years but I got fed up with them after they continuously questioned what I told them of a failing disk in the RAID set of one of the servers and then I moved.

But for a VPS, well, can they at all fail? It's still yet to be seen, I guess... :)

Good luck!
 
OptiBiz1 said:
Server4You, but if you are unlucky with what support technician you are in contact with, they suck *ss. I gotta tell you that.
I was their customer on dedicated servers for six years but I got fed up with them after they continuously questioned what I told them of a failing disk in the RAID set of one of the servers and then I moved.

But for a VPS, well, can they at all fail? It's still yet to be seen, I guess... :)

Good luck!
Click to expand...
The lower level vps have such ridiculous limitations, that they make no sense to be used. Then they try to upsell you like hell. I tried them for a few days, then moved away, years ago :(
But their backup snapshot feature is great (they must be using r1soft or so with own interface).
 
You'll want to look and see where the most time is spent, mailwizz shows timestamps for all actions so you can compare and see how much time is spent for each action. When you start observing a pattern, you'll get the culprit. Make sure you disable pcntl if you want to get clear readings.
 
You must log in or register to reply here.
Back
Top