Every time somebody logs in the pam_motd module runs all the scripts in
/etc/update-motd.d/. One of these scripts is
98-fsck-at-reboot which runs
/usr/lib/update-notifier/update-motd-fsck-at-reboot. In essence this script, for each ext filesystem, checks the mount count against the maximal mount count and the last fsck against the check interval to determine if a filesystem needs to be fscked, and, if so, outputs text that gets added to
When a system boots, upstart examines
/etc/init/mountall.conf and runs
mountall, which runs fsck on the filesystems if the conditions stated in the previous paragraph are met.
As you can see these goings-on are not triggered by detected errors and so probably have nothing to do with your database connectivity issues.
Most likely you are exhausting a resource (I'm guessing memory) on your micro instance and a reboot causes resource consumption to be reset. If you don't wish to troubleshoot it any further I'd upgrade to at least a small instance.