Problem
Your system is experiencing many errors in our Jira log... nearly 300k in 2 days showing
Dangerous use of multiple connections: replaced => count=0; marks=[-1-0]; pool=6/20
This causes the available db connection pool to get exhausted and is causing the entire Jira instance to crash.
Solution
Unfortunately, this problem has no clear solution, however Atlassian has confirmed a bug, see JRASERVER-45794 for workaround suggested by Atlassian.
Otherwise, it may be that problem happens due to dashboard left unattended, when session expires. You may try to remove all add-on gadgets from all dashboards, if it's acceptable. See https://bitbucket.org/tempo-io/servertimesheet/wiki/Usage_Tracking for how to find them.
Stopping JIRA service, rebooting the DB server, then starting JIRA service might help.
A possible workaround is to consider using Time Reports for Jira Server instead, it has quite same UI, but different backend, so it does not have such problems. See also Comparison Matrix for feature differences.
See also issue#1066 and contact Atlassian Support for help.