Follow up to Outlook issue
The issue appears to be related to zimbraHttpNumThreads which was set to 250. We have increased this to 500 and performed a restart of the Zimbra service on the server concerned. We have also implemented an additional MRTG graph to monitor this metric. Our servers are performance tuned in accordance with Zimbra's best practise guidelines, although this specific setting is not manditory and the guidelines are slightly ambiguous as to what value is most appropriate.