Thursday, November 20, 2008
Inbound email
When it rains it pours.
We have located another issue with one of the primary MX cluster members, and the way another member uses DNS - these occured over night (19th-20th November).
These have been addressed, and we are seeing mail processing in at an expected rate given the time of day.
Please do keep in mind the way the mail RFC's work, and that unless you have asked the admin at the other end to force a flush, mails retry with a logarithmic delay, so be patient, and new mail may arrive before ones sent less recently. This is the way it is meant to work to assist with resilience and spread the load.
We are sorry for the delays, and are happy to say that there are now no issues.
We have located another issue with one of the primary MX cluster members, and the way another member uses DNS - these occured over night (19th-20th November).
These have been addressed, and we are seeing mail processing in at an expected rate given the time of day.
Please do keep in mind the way the mail RFC's work, and that unless you have asked the admin at the other end to force a flush, mails retry with a logarithmic delay, so be patient, and new mail may arrive before ones sent less recently. This is the way it is meant to work to assist with resilience and spread the load.
We are sorry for the delays, and are happy to say that there are now no issues.
SMTP
Outbound email is currently being delayed.
Client abuse has resulted in a huge spam list that we have successfully cleared from the queue and contacted / blacklisted those involved.
Trying to clear the backlog appears to be being hampered as - quite logically users are sending themselves test messages, or mails that have not arrived have been resent. While this may make sense on the scale of one person, upscale this a couple of hundred times, and there is a fair amount of mail happening that doesnt need to.
We are aware of the issue, and we are processing the queue as well as we are able - once the queue has returned below 8K mails, we are confident it will start to recover of its own accord. Bear with us, and we will update you once we have the green light.
[update 0955] Blaming the tail end of an issue yesterday for this mornings issues quickly became a non starter. We have located the issues and made good. I have posted a new article to cover what happened.
Client abuse has resulted in a huge spam list that we have successfully cleared from the queue and contacted / blacklisted those involved.
Trying to clear the backlog appears to be being hampered as - quite logically users are sending themselves test messages, or mails that have not arrived have been resent. While this may make sense on the scale of one person, upscale this a couple of hundred times, and there is a fair amount of mail happening that doesnt need to.
We are aware of the issue, and we are processing the queue as well as we are able - once the queue has returned below 8K mails, we are confident it will start to recover of its own accord. Bear with us, and we will update you once we have the green light.
[update 0955] Blaming the tail end of an issue yesterday for this mornings issues quickly became a non starter. We have located the issues and made good. I have posted a new article to cover what happened.
