Questions: SNMP, alert config, and a couple potential bugs
I am currently evaluating Applications Manager (build 7410 on WinXP) in a toy environment using the free version, and I have a few questions. So far I have the following monitors configured:
- Linux server monitor
- SNMP monitor
- SNMP trap listener
- Java Runtime monitor
- JMX Applications monitor
1. I've read that hierarchical structuring of monitor groups is now supported but I don't see how to do that. Is that functionality only available in the non-free versions?
2. My SNMP monitor has three SNMP table attributes configured for polling and display based on the UCD-SNMP-MIB: laEntry (7 attributes), dskEntry (12 attributes), and prTable (9 attributes). Each table is displayed in Applications Manager in a separate GUI table. The second and third tables displayed have the title "SNMP Table Name : dskEntry" and "SNMP Table Name : prTable". The first table (regardless of which SNMP table it is) always has its name prefixed by "MBean Name" (for example, "MBean Name : laEntry"). Is this labelling of an SNMP table as an MBean a bug?
3. The alert configuration mechanism seems to allow alerts to be configured based on one attribute only. Perhaps I'm thinking about this incorrectly, but I'd like to be able to configure an alert based on multiple attributes. For example, the prTable (process table) SNMP table mentioned above has prCount (process count) and prNames (process names). I currently have two rows in the table monitoring the number of sshd and ntpd processes on the box. I have an alert configured to take an action if the prCount is <= zero. What I would like is to have an action occur if the prCount is <= zero and the prNames in the row is "ntpd". However, this doesn't seem possible in the Applications Manager GUI. Is this type of complex alert possible?
4. In the threshold definition in item 3 above (based on prCount) I also have an action associated with a Clear Severity. If the prCount is >= 1, then the alert is cleared and an action is taken (email). The issue is that the SNMP monitor is fetching two rows in the prTable (one for ntpd and one for sshd). If ntpd is down and prCount is 0 a critical alert occurs. At the next poll (when ntpd is still down), the alert is cleared because the results for the sshd row (prCount of 4) cause the alert to clear. The result is an alert that toggles between critical and clear on each poll (thus sending lots of emails). This is either a side effect of the lack of non-complex alerts cited in item 3 above, or its a bug in that the alerts seem to assume only one row of SNMP data is going to come back for a table and base decisions off the first row returned (which seems to toggle). Any insight into this is appreciated.
5. When I set up a Trap Listener, the monitor associated with it (as seen in Action-based emails) is the Trap Name field specified when the trap listener was created. When I renamed the trap listener through the Trap Listener Details screen, the new name is used when viewing trap listeners in the GUI, but the associated Monitor name (seen in Action-based emails) is still the old name. Is this a bug?
Thanks...
Brett
- Linux server monitor
- SNMP monitor
- SNMP trap listener
- Java Runtime monitor
- JMX Applications monitor
1. I've read that hierarchical structuring of monitor groups is now supported but I don't see how to do that. Is that functionality only available in the non-free versions?
2. My SNMP monitor has three SNMP table attributes configured for polling and display based on the UCD-SNMP-MIB: laEntry (7 attributes), dskEntry (12 attributes), and prTable (9 attributes). Each table is displayed in Applications Manager in a separate GUI table. The second and third tables displayed have the title "SNMP Table Name : dskEntry" and "SNMP Table Name : prTable". The first table (regardless of which SNMP table it is) always has its name prefixed by "MBean Name" (for example, "MBean Name : laEntry"). Is this labelling of an SNMP table as an MBean a bug?
3. The alert configuration mechanism seems to allow alerts to be configured based on one attribute only. Perhaps I'm thinking about this incorrectly, but I'd like to be able to configure an alert based on multiple attributes. For example, the prTable (process table) SNMP table mentioned above has prCount (process count) and prNames (process names). I currently have two rows in the table monitoring the number of sshd and ntpd processes on the box. I have an alert configured to take an action if the prCount is <= zero. What I would like is to have an action occur if the prCount is <= zero and the prNames in the row is "ntpd". However, this doesn't seem possible in the Applications Manager GUI. Is this type of complex alert possible?
4. In the threshold definition in item 3 above (based on prCount) I also have an action associated with a Clear Severity. If the prCount is >= 1, then the alert is cleared and an action is taken (email). The issue is that the SNMP monitor is fetching two rows in the prTable (one for ntpd and one for sshd). If ntpd is down and prCount is 0 a critical alert occurs. At the next poll (when ntpd is still down), the alert is cleared because the results for the sshd row (prCount of 4) cause the alert to clear. The result is an alert that toggles between critical and clear on each poll (thus sending lots of emails). This is either a side effect of the lack of non-complex alerts cited in item 3 above, or its a bug in that the alerts seem to assume only one row of SNMP data is going to come back for a table and base decisions off the first row returned (which seems to toggle). Any insight into this is appreciated.
5. When I set up a Trap Listener, the monitor associated with it (as seen in Action-based emails) is the Trap Name field specified when the trap listener was created. When I renamed the trap listener through the Trap Listener Details screen, the new name is used when viewing trap listeners in the GUI, but the associated Monitor name (seen in Action-based emails) is still the old name. Is this a bug?
Thanks...
Brett