But if im right. When i run that command it fires off 500 at a time correct? So I will have to keep doing this until my 120k emails are up? That sounds like alot of time.
High CPU Useage
- Thread starter Eric535
- Start date
But if im right. When i run that command it fires off 500 at a time correct? So I will have to keep doing this until my 120k emails are up? That sounds like alot of time.
twisted1919
Administrator
Staff memberThe idea is that you do this a few times till you spot where the bottle neck is. You won't be sending the entire campaign like that. When you are done debugging, you resume normal sending.
I wonder If I can use one of my domains that are sitting around and make a catch-all inbox with 1000 emails matching that domain?
How do I resume after? Do i close the terminal, re-enable the cron job and it will continue automatically?
twisted1919
Administrator
Staff memberThe timings, i explained above 
The timings, i explained above
Got it. What are the chances of the bottle neck happening in the middle of the list? LOL
twisted1919
Administrator
Staff memberAlmost 0. You will see per subscriber basis what's going on.
I dont want to test on real emaill accounts. Can i just to the catch-all thing i mentioned earlier? Or could the problem be the list itself?
twisted1919
Administrator
Staff memberYou should test as close as real world as possible.
frm.mwz
Well-Known Member
You can easily do a catch all even on a sub domain and filter them nicely to later see when sent, when arrived, etc.
Once you do manual sending from cli you will really enjoy the analysis the output allows.
You can pipe it into a text file or ask the terminal to copy the output and then paste it into a good text editor for further work on the output.
Just do it, and you will get the hang of it
You can easily do a catch all even on a sub domain and filter them nicely to later see when sent, when arrived, etc.
Once you do manual sending from cli you will really enjoy the analysis the output allows.
You can pipe it into a text file or ask the terminal to copy the output and then paste it into a good text editor for further work on the output.
Just do it, and you will get the hang of it
@frm
Didnt see this until now. I guess I can do it on a subdomain also. If I wanted to I could also make a bot to open or click through the emails to see some more stats.
Thanks
You can easily do a catch all even on a sub domain and filter them nicely to later see when sent, when arrived, etc.
Once you do manual sending from cli you will really enjoy the analysis the output allows.
You can pipe it into a text file or ask the terminal to copy the output and then paste it into a good text editor for further work on the output.
Just do it, and you will get the hang of it
@twisted1919 I updated to 1.51 hoping it might of fixed the problem but it didnt. I will do the debugging test next week when I send off another campaign. But as you can see by the snapshot the CPU usage goes out of control at 11:00am. At 11:00am is when I send the BIG list of 120k emails and thats what makes the cpu go funky. Before 11:00am I send out a smaller list (about 4k) which doesnt get the CPU up that high and it takes 7 mins to send. Hmmmmmmm
frm.mwz
Well-Known Member
Then it is most likely server setup or settings (but not mwz).
I have two other server setups with MWZ the same way with less hardware and those two email faster and dont have high CPU. The only difference is that the other two i use sparkpost and this one I use amazon ses for sending the emails
frm.mwz
Well-Known Member
...same server setup, but different hardware...and you want the same/better results?
...settings
That is why the prev post was:
If you want to progress this, it might be useful to become more specific, e.g. post some of your cron and DS settings for the new server.
Attached are the snapshots for my (Linode 4096) Delivery Server and snapshots of the cron settings for both the linode 2048 and the linode 4096. As you can see the smaller 2048 does in fact have different settings. I dont understand why the bigger linode 4096 would be having high CPU issues if the settings are less aggressive than the smaller ones. I dont have any CPU issues with the smaller ones.
See snapshots below.
Amazon Delivery Server Settings for the Server in question (Linode 4096)
Cron Settings for the Server in question (Linode 4096)
Cron Settings for the Server in question (Linode 2048)
frm.mwz
Well-Known Member
Why not try the marked cron settings from Linode 2048 (2500/m) also at Linode 4096 (where settings as posted make no sense, with 5x100 to send of each of 5 camps, but only 50 to send at once, and 100/m (when 500 ready at least, probly 2500) but change to same DS after 200, basically all a brake, which is what you experience)?
Vpul Shah
Well-Known Member
@twisted1919
So after upgrading servers and doing everything else mentioned in this thread, I narrowed down the CPU usage and speed to a list that I send to which is about 120k emails. Whenever I send to that particular list, it takes around 12+ hours and thats when I get the high cpu usage warnings. I have another list which is half the size and only takes 2 hours and never gets high CPU warnings. Why would a particular list slow down my server?
So after upgrading servers and doing everything else mentioned in this thread, I narrowed down the CPU usage and speed to a list that I send to which is about 120k emails. Whenever I send to that particular list, it takes around 12+ hours and thats when I get the high cpu usage warnings. I have another list which is half the size and only takes 2 hours and never gets high CPU warnings. Why would a particular list slow down my server?