Email sending issue again (Was: URGENT)
-
Everything seems okay... I do not know why it automatically refused connection. Each time this happens I try to change the domain and the IP. But I cannot seem to understand why this keeps repeating itself.
I need to know so I can understand the way forward on this. I have sent the logs as requested
-
Everything seems okay... I do not know why it automatically refused connection. Each time this happens I try to change the domain and the IP. But I cannot seem to understand why this keeps repeating itself.
I need to know so I can understand the way forward on this. I have sent the logs as requested
@mmtrade thanks, received . So the issue is that Haraka (the SMTP server) keeps crashing . I suspect this is because the mail queue is corrupt. You can clear it using the commands at https://docs.cloudron.io/email/#haraka-queue-corruption . Can you run the commands there via SSH and let us know if it works after that ?
-
@mmtrade Are you already on Cloudron 8.3 ? It helps if you have the latest Haraka so that we can debug productively .
If this happens again, can you send a mail to support@cloudron.io and we have to debug it in your instance. It's some Haraka bug . We have to make an issue upstream and possibly provide a patch.
-
-
-
@mmtrade we only provide best-effort support through the forum. If you need more timely and direct support, we have a subscription plan for this.
-
-
I have made a new build of mail container.
IMPORTANT: Do not apply this fix unless you are affected by this . The new mail container passes tests but it also contains a new Haraka release.
- Edit
/home/yellowtent/box/src/infra_version.js
. Change the mail container image toregistry.docker.com/cloudron/mail:3.16.1@sha256:900b7f55a74a289568c501cd873c24738d424e24f005c3c49cf76a26c296ec22
systemctl restart box
- Edit
-
It’s not what you communicated but how.
And I believe it‘s far more commendable that @girish kept his patience, analyzed your particular problem (no one else reported something similar), reported it upstream, and merged the fix himself, all within hours. And all without an SLA.
So credit where credit is due.