Port checks and Mobilink server

When we turned on our Mobilink server for the first time, we were baffled by the fact that we repeatedly got network errors in our log.

E. 2008-12-09 19:01:10. <1> [-10030] A network read failed.
Unable to read data from the remote client
E. 2008-12-09 19:01:10. <1> [-10091] This connection will be
abandoned due to previous errors
E. 2008-12-09 19:06:10. <2> [-10030] A network read failed.
Unable to read data from the remote client
E. 2008-12-09 19:06:10. <2> [-10091] This connection will be
abandoned due to previous errors
E. 2008-12-09 19:11:10. <3> [-10030] A network read failed.
Unable to read data from the remote client
E. 2008-12-09 19:11:10. <3> [-10091] This connection will be
abandoned due to previous errors

We thought perhaps one of the stores had some nasty network problems, dropping packets. As we looked closer, we saw that these errors were occurring exactly every five minutes, even during a successful sync. However, our sync schedule is every fifteen minutes.

After talking to our sysadmin, he let us know that every five minutes the server was being checked for the "liveness" of that port by Nagios. However, the timestamps on the check log and the Mobilink log didn't match up. As a test, we spaced those checks out to every ten minutes, and then the errors occurred every ten minutes.

It turns out that when Mobilink got a tcp/ip signal on its port, it was trying to initiate a sync even though the signal coming through wasn't a "please let me share" kinda signal. That was why the numbers in the <>s were incrementing by one every time-- it's a synchronization autoincrement. Mobilink was just taking a second to time out, so the timestamps didn't match.

So, if you're getting these strange errors, it may not be a network problem at all, but something non-Mobilink trying to hit your port.