2 Replies Latest reply on Aug 19, 2019 1:00 PM by EvanEverett

    Broker Agent service causing SQL blocking and deadlocks?

    ooajala Enthusiast

      Hi Community,


      I am not sure if anyone can help me with this before I put in an official SR with vmware. So we have a couple of connection servers that have the broker agent installed which are our View Pods.


      We have had issues with the broker agent service using most of the RAM and in addition, the database team have complained about SQL deadlocks and blocking occurring. Those blocking are gone when I stop the broker agent service. I know this cannot be the fix because I definitely need the service on to record Horizon connected sessions metrics etc.



      The below is on the SQL database server managing one of the connection servers.



      Is anyone experiencing this kind of issue?


      Please advise.


      Thanks in advance.

        • 1. Re: Broker Agent service causing SQL blocking and deadlocks?
          EvanEverett Novice

          Hello ooajala,


          Were you ever able to get an answer on this? It would seem our organization is running into a similar issue...





          • 2. Re: Broker Agent service causing SQL blocking and deadlocks?
            EvanEverett Novice

            In case anyone else runs into this issue, this is a bug with Horizon View. Per VMware Support:


            As covered, this is a known issue identified on the Horizon View side.  It is targeted for the 7.10 release of Horizon view (tentatively Q3, but subject to change).

            The only workaround is to lower the Event Settings configured in View Configuration > Event Configuration tab to as low as you can for the interval seen there.


            As a side note and word of caution, be careful when you adjust the Event Configuration from 3 Months (Max) to 1 Week (Min) retention policy. The SQL DB is purged and causes extreme congestion with IO at the disk level. I also tested going from 3 Months to 2 Months (next lowest retention policy) and still had the issues of maxing out the SQL DB I/O. The solution for us was to create a new SQL DB, change the event retention to one week, and then point to the new SQL Database on the events.