using the new pcntl sending with PMTA

Jason

Member
hey,

so keeping most of the setup the same as i had, whilst changing back the cron to match Mailwizz, and enabling parallel sending with pcntl in cron section in MW.

this does not now work. PMTA does not receive anything

All mdp-go info including directory is still on my server, do i need to get rid of it? I killed the mdp daemon on cmd line.

any advice on what i should do/test? I should add im pretty sure pcntl is installed.

thanks
 

Jason

Member
ok so i forgot i would have to kill the daemon for parallel sending originally changed for mdp go

so running the following i got;

[root@platform /]# ps aux | grep send-parallel-campaigns
root 11936 0.0 0.0 103364 1928 pts/0 S+ 13:24 0:00 grep send-parallel-campaigns

[root@platform /]# killall -15 send-parallel-campaigns
send-parallel-campaigns: no process killed

tried hitting the different number ids to kill that process, but wouldnt do it? I think im on the right track here?
 

Jason

Member
so that resulted in;

/usr/bin/php-cli -q /home/admin/web/platform.xyz/public_html/apps/console/console.php send-campaigns --verbose=1
-bash: /usr/bin/php-cli: No such file or directory
[root@platform ~]# /usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php send-campaigns --verbose=1
[2016-02-18 14:57:33] - Loading 5 campaigns, starting with offset 0...
[2016-02-18 14:57:33] - No campaign found, stopping.

my cron jobs currently, as specified in my CP are;

/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php send-campaigns >/dev/null 2>&1
/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php daily >/dev/null 2>&1
/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php process-delivery-and-bounce-log >/dev/null 2>&1
/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php feedback-loop-handler >/dev/null 2>&1
/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php bounce-handler >/dev/null 2>&1
/usr/bin/php -q /home/admin/web/platform.xyz/public_html/apps/console/console.php send-transactional-emails >/dev/null 2>&1

for some reason i can't see them on the command line, may be limitations of my knowledge showing there.. sorry

thanks
 

Jason

Member
okay, so this is occuring when i try and do a confirmation email for a pickup server. Usually it shoots through the specified server.

although just a thought.. I may have a old server specified in the config.json file in the mdp-go folder, would this affect this process in anyway? I was thinking this old process for MDP-GO would now be bypassed completely. But how does pcntl know what IP to shoot to through a pickup server... unless i should be using SMTP instead?

also;
~]# crontab -l
no crontab for root
 

Jason

Member
okay well i've just tested the SMTP server in MW to PMTA (which worked all ok) and that connected okay, so have tested the spooling speed.

114k per 60 mins, not bad! And with slightly dropping out once which is great, constant problem i had with MDP go.

just want to confirm this is the process i am meant to use now for parallel sending with pcntl? I take it, it is because normal speed of this previously via SMTP was terrible. So cool stuff!

thanks
 

Jesse James

Member
idk, for me its working fine i guess!
Please check the attached files.
Screen_Shot_2016_02_18_at_11_51_14_PM.png

Screen_Shot_2016_02_18_at_11_46_54_PM.png
 

Jason

Member
thats cool!

so i left it to test spooling overnight, and the total volume of the list is 280k, somehow a total of 433k has spooled into PMTA...

I can only assume it sent them in twice sort of :S, how did this happen?? Fairly sure no other send daemon is running
 

bidorbuy

Member
okay well i've just tested the SMTP server in MW to PMTA (which worked all ok) and that connected okay, so have tested the spooling speed.

114k per 60 mins, not bad! And with slightly dropping out once which is great, constant problem i had with MDP go.

just want to confirm this is the process i am meant to use now for parallel sending with pcntl? I take it, it is because normal speed of this previously via SMTP was terrible. So cool stuff!

thanks

Do you mind sharing your full delivery settings - i.e. like this:
upload_2016-2-22_8-31-9.png
 

bidorbuy

Member
not at all, here you go;

View attachment 1457

Thank you. Have you done a cross check between what MailWizz says it has sent and how many mails actually arrived in Port25? I noticed that the current version of SwiftMailer shipped with MailWizz caused us to drop almost 80% of mails (i.e. MailWizz would report mails being transmitted, but when checking Port25 most of them never arrived - (Example: Mailwizz=82K mails / Port25=20K). To me this seemed to be a networking issue within SwiftMailer and also affected local relay into Postfix.

@twisted1919 then provided a SwiftMailer upgrade which solved that issue. I have not yet tried PCNTL as other people had PCNTL issues. If you have the same issue, your throughput stats are then wrong.
 

Jason

Member
hmm, no i havent. Please can you advise the best way to go about a quick check of this and fix?

not tried to upgrade swiftmailer before
 

bidorbuy

Member
hmm, no i havent. Please can you advise the best way to go about a quick check of this and fix?

not tried to upgrade swiftmailer before

In our setup we write accounting files on Port25 (via setting "act-file"). In MailWizz we use "campaign.bounce@domain.com" for bounce addresses and then on Port25 we do a simple grep with a count:
Code:
grep -i "campaign.bounce" /var/log/pmta/pmta-bounce-acct-$(date +'%Y-%m-%d')-0000.csv  | wc -l

The above command will give you a count of mails received by Port25. In our test campaign for today we sent via MailWizz to 2544 subscribers and the above count on Port25 returned 2736 (the count will be higher as we did some smaller test campaigns, but Port25 should never be less).
 
Top