Posts Tagged ‘health center’

DB2 notifications

February 16, 2012

Notifications are used in two parts of DB2 – Health Center and Task Center. Configuration is pretty simple. Go to Control Center – Task Center – Tools – Contacts and click on SMTP Server button. Fill here the FQDN of your mail server.  Then add contacts. After that you can use them during scheduled task configuration on Notifications tab.

To configure Health Center notifications go to Control Center – Health Center – Health Center – Configure – Alert Notification. In Configure Health Alert Notification window select instance and add contacts to Health Notification Contact List.

It may look like this is it. However, there is one issue we ran into. Sometimes DB2 sends emails using just a host name, without the domain part, like db2admin@sqldb2. All modern servers won’t allow sender address not in FQDN format. So we had to add DB2 server IP into white list on our Postfix mail server using check_client_access parameter.

DB2 Monitor Heap Utilization alert

February 16, 2012

Several times a week we receive alerts from DB2 Health Center concerning utilization of heap memory consumed by Health Monitor. Here is the message from logs:

ADM10500E  Health indicator “Monitor Heap Utilization” (“db2.mon_heap_util”) breached the “upper” alarm threshold of “95 %” with value “200 %” on “instance” “DB2”.  Calculation: “((db2.mon_heap_cur_size/db2.mon_heap_max_size)*100);” = “((655360 / 327680) * 100)” = “200 %”.  History (Timestamp, Value, Formula): “()”

Even though MON_HEAP_SZ parameter of our DB2 instance is configured to be managed automatically alert still shows up. As it turned out it’s a minor DB2 bug. Which is described here. Short explanation of this bug follows:

Health Monitor is sending Alert to db2 user while Monitor Heap is set to Automatic.

So if in configuration parameters you see “AUTOMATIC” in Value column for MON_HEAP_SZ then you can safely ignore this alert. However, it’s not a good idea to have unresolved issues in Health Center. Besides, it’s rather annoying since it shows up quite regularly and bring our attention to it. The solution for this problem is to switch off monitoring of monitor these parameters by:

db2 update alert cfg for database manager using db2.mon_heap_util set THRESHOLDSCHECKED NO
db2 update alert cfg for databases using db.db_heap_util set THRESHOLDSCHECKED NO

What’s peculiar about this error is that according to IBM it was fixed in 9.5 and we work on 9.7 (regression bug?). On top of that, there is another bug JR31509 described here which is connected with the previous one. Short explanation:

The health indicator calculates the alarm threshold using db2.mon_heap_max_size, but this max size may not be increased even though the MON_HEAP_SZ is configured as AUTOMATIC.

That basically means that AUTOMATIC may not work properly and we might have memory issues in future. But I’ll keep my fingers crossed.