8 Replies Latest reply on May 23, 2013 5:10 AM by michael.russell.ubm

    Advice on keeping emails from your Jive community out of Postini's filters


      We've had our (hosted) Jive system up and running for a long time, but are just now getting deeper into figuring out why not all of the emails coming from it are getting through to employees. I'm going to have one of our IT guys (Michael Russell) weigh in here as well, since he can speak to all of the technical details.


      The over-arching theme from him is that he can't find anything consistent to help him create appropriate rules so that all the emails get through. The messages don't have consistency. And I can explain some of that:

      1. Some are system messages, so they have a reply-to for the email address defined by us. This way, the system administrator gets any replies to those emails.
      2. Some are messages that can be replied to, where the response will get posted back to the original thread. Those need a unique reply-to so the original thread can be matched.
      3. Some are messages where the response is allowed to go directly back to the originator. Again, this is due to the type of message that it is.


      So I understand some of the lack of consistency, and that these are important features from the enduser experience.


      However, I'm at a loss as to what to tell the folks who are trying to write Postini rules such that the Jive emails don't randomly get caught in the SPAM filters.


      What advice is there from this group on the subject? What have you been able to write to reduce or eliminate this issue for your employees?

        • Re: Advice on keeping emails from your Jive community out of Postini's filters


          We figured this out pretty quickly for our internal hosted instance.  I can get the details on what we did in our environment if that would be helpful.

          • Re: Advice on keeping emails from your Jive community out of Postini's filters

            Hi, I'll overview what we've done and our day to day business (distinct from my role) is business to business marketing so we're very aware of the problems caused my system generated messages and mass mailing.


            We use SPF. Jive's MTAs are part of these records so they are recognised as approved senders for @ubm.com addresses. (If SPF isn’t in use we define all MTAs as approved senders)
            We use conventional sender white lists (although these are becoming less effective for bulk mailers).
            We can search for values in messages which can trigger an allow rule (Content filters).


            There are a number of rules that can be triggered automatically in Postini, the most frequent for Jive originated email is "Bulk". The sender is detected as sending an abnormally high volume and its messages are quarantined. Email classified as "Bulk" doesn't care about SPF or approved senders, the message won’t get delivered.


            We can use a content filter which will ignore all spam rules. It's a high impact solution because anything that matches the rule will get delivered. A spammer could start filling their messages with stuff in the hope they trigger these rules in which case the message would be delivered. When we define a content filter we target a specific part of the message and look for something very specific within that. We must avoid false positives.


            We've had to start using these content filters for Jive emails. We need to use consistent triggers so we avoid targeting IP addresses (because there are many of them, not often sequential and often change), DNS names (same reason as IP addresses), from addresses (far too generic and likely to allow too many), anything generic (like subjects, daemon names).


            We would normally use data from the headers and after looking at a few example messages we thought we had something. As we've watched our quarantine lists over time we're still seeing things being captured and the annoying thing is that the headers are often different. Some have a specific "Return-Path:" value, some have a "X-Jive-Office:" field and some have no specific header tags at all.


            I'm surprised that given the number of different techniques being used to send mail to us from Jive (sending as, bulk mailing, different reply to's, many MTA, multiple systems, etc) there isn’t a specific header value that's not going to change we can use to target a rule at and get these mails through. I can see that the system is capable of inserting a x-header value but it snot on all messages from Jive.


            Can you look at putting something in the header on all emails from your system so we can capture them and bypass our spam rules?


            • Re: Advice on keeping emails from your Jive community out of Postini's filters

              We've had our IT department white list various Jive email addresses which takes care of email notifications.


              However, we've had a lot of problems with @mentions and direct messages getting stuck in Postini. These types of messages do not come from Jive; they come from firstname.lastname@company.com -- in other words, our own domain. The only way around this is to white list individual names, which does work, but that means we are constantly missing @mentions or DMs from any "new" contacts whom we have not yet had a chance to add to the safe sender list. This we haven't yet been able to figure out.

              • Re: Advice on keeping emails from your Jive community out of Postini's filters

                So I'm feeling like there is still not a complete answer/resolution to this. Maybe someone in Jive Developers might have suggestions? Or someone in Jive like Tyler Whitesides or Arnold Benson? Would love to get to the bottom of this, as I have several people actively complaining at the moment.