As of 1 AM EST / 7 AM GMT the monthly global maintenance cycle has been completed. Please allow up to 2 hours for routes to converge and new networks to be announced.
Among updates, features and bug fixes are:
- Provisioning for the new Los Angeles, CA and Chantilly, VA networks.
- Expansion of LiveArchive network.
- Fix for the ExchangeDefender routing network priority hack (for low bandwidth mail servers)
- Fix for Australia/Pacific routing group (* extending smtp_greet from 30 to 120 seconds to compensate for trans-Pacific)
- Extended nameserver network to improve load balancing.
- Provisioning for the new RR routing strategy for international customers with complex compliance issues (ex: “route all mail through UK, route through EU in case of network outages, never route outside of EU”)
It will take at most two hours for the routes to converge, during this time you may see delayed messages or deliveries out of sequence as the new paths are being distributed. You also may see delays in extended-outage-queues we use for disaster recovery (ETRN/queueing/spooling) mechanisms which are affected by all of the above but primarily the name server expansion.
These efforts are critical to the continued growth and reliability of the ExchangeDefender network. No downtime or outage has occurred.
We experienced a brief power failure at 10:20 PM Central in our Dallas 3 data center. The outage affected roughly 20 servers and was caused by a failure in a PDU (power strip) that tripped. Outage affected a very small section of the ExchangeDefender network which is fully redundant and was not affected at all by this outage. As a precaution, we have taken the affected nodes out of the scanning pool until their hourly reload of network configuration at which point they will resume normal operations.
ExchangeDefender uptime, availability and load were not affected as this is a very low activity time window and a very small portion of the scanning network. As our storage arrays are not on the same PDU as the scanning nodes, there was no interruption or delay in scanning service.
We are currently running an experimental delivery process on ExchangeDefender mail that has been queued on ExchangeDefender delivery queue for over an hour. If our management system indicates that the target server is up and the message is continuously deferred/rejected, the message will be handed off to another delivery agent that will attempt to stream it through a different process.
To review, if the following conditions are met:
- Recipients mail server is up and running
- Recipients SMTP port is available and accepting connections (no greylisting).
- Message is older than an hour and smaller than 100 Mb.
.. our system will attempt to flush the message through a different system.
These messages may be malformed in some way and the recipients server never would have received them directly from a third party server. However, with ExchangeDefender in the middle, we believe we can attempt to compensate for the problems in the message.
What to watch out for:
Our redelivery queue currently stands at 14 days. It is possible that you will see messages that are older than just a few days.
Why are we doing this:
We have decided to get to the bottom of the sporadic issues that have been reported through the years and build a concise, managed system that monitors for clients server problems so support requests can have a more timely and adequate response. We will extend this Managed ExchangeDefender service as we go along.