| Title: | DECmcc user notes file. Does not replace IPMT. |
| Notice: | Use IPMT for problems. Newsletter location in note 6187 |
| Moderator: | TAEC::BEROUD |
| Created: | Mon Aug 21 1989 |
| Last Modified: | Wed Jun 04 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6497 |
| Total number of notes: | 27359 |
We have a problem running total error rate statistics on a Vitalink bridged network. The rules are looking for Total Error Rates exceeding 1% for the Ethernet and v35 ports. The rules fail consistently with Total error rates in the order of 4% - 10%. On investigation the port error counters are not increasing apart from the ifindiscards (Which I believe is the frames filtered by the bridge). The Vitalink counters show that the incrementing variables are :- frames transmitted frames received multicast transmitted multicast received recieve discarded The Polycenter Performance analyzer SNMP statistics show the formular to be :- Total error = (ifindiscards+ifinerrors+ifinunknownprotos) packet % +(ifoutdiscards+ifouterrors) ----------------------------------------- ((ifinUcastpkts+ifinUcastpkts+ifindiscards +ifinerrors+ifunknownprotos) +(ifoutUcastpkts+ifoutNucastpkts+ifouterrors)) It would imply that the ifindiscards counter is causing the high result, so is the inclusion of ifindiscards valid for a bridge port where discarding frames is a normal function. For a router port the ifindiscards would not increase in normal operation. The Lanbridge algorithm does not use frames discarded, so is the performance analyzer getting it wrong for a vitalink bridge using SNMP error statistics. Can anyone help Thanks Mat Evans
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 5787.1 | Documented behavior; you *can* code around it. | IMDOWN::MCPHERSON | Interstate Hwys in Hawaii? | Thu Dec 16 1993 13:53 | 16 |
LANbridge algorithms are specific to lanbridges.
SNMP algorithms apply to ALL SNMP obects that return MIB2 attributes.
You cannot tailor the SNMP PA algorithms.
You cannot tailor the LANbridge algorithms.
In summary: there is nothing that can be done to alleviate your
problem, as stated.
Your only options are
1. change the statistic that you are looking for, or
2. write your own FM that either
a. gathers the exact variables that you want from the SNMP AM or
b. calls the PA FM for the SNMP objects in question & 'tailors' the
output of the PA equations, based on your own tables. (may
actually be easier to code than option b.)
/doug
| |||||