MFD : Frequently Asked Questions

One of the stengths of the MIBs for Dummies configuration file is also one of its weakenesses: flexibility. There are many different customizations that can be made to the code. Hopefully this document can help provide some guidance on when you might want to choose one option over another.

 

---

 

mib2c

 

---

 

How do I customize the generated code?

There are two ways to control how mib2c generates code. One is via command line arguments, and the other is via configuration files. Both methods are covered on the MFD options page.

 

---

 

How can I re-generated code for my table without losing my changes?

There is no method that will gurantee a smooth transition, but if the code templates have not changed to radically, it is a fairly simple process. There is even a shell script to automate the process. Details can be found on the mib2c-update page.

 

---

 

Data contexts

 

---

 

When should I initialize the data context?

When a request is received for a particular table, the indexes for the table must be loaded to determine a row can be found to respond to the request. Initializing the data context, however, can be delayed until the data for a particular row is needed. So, how do you decide when to initialize the data context?

That depends on several issues, including how fast you can access your data, how much data you have, whether or not you are caching your data, and how you expect your data to be accessed.

Initialize the data context while the indexes are being initialized if:

• you have a small data set such that initializing the data context has a negligible impact on load time
• your data changes rapidly and you want requests within the cache timeout to use data from the same timeframe or sample set. (eg during a snmpwalk of the table).

Delay initialization of the data context until the data is needed if:

• you do not care about data from different timeframes or sample sets being used for requests (eg during a snmpwalk of the table).
• you always want the freshest, most recent data to answer requests
• you have an extremely large data set, and initializing the data context impacts performance, or requires excessive amounts of memory

 

---

 

Do I really have to load all the indexes?

The idea behind the MFD templates is to require as little SNMP knowledge as possible. However, if you have the SNMP knowleged and want to make some optimizations, we will point you in the right direction. We might not be as sympathetic when you run into problems, though.

SNMP requests fall into one of two categories: requests that specify a specific object (SnmpGet, SnmpSet), and those that don't (SnmpGetNext, SnmpGetBulk).

...

 

---

rstory at freesnmp dot com
Version $Revision$