Cabletron Systems DMS-100 manual In-service diagnostics, In-service leaky bucket audit

Chapter 4: EIU maintenance 105

or the MAP display. The TST command in MAPCI also executes the same diagnostics when the EIU is ManB.

In-service diagnostics

The EIU changes its state to SysB state from InSv or IsTb if a serious hardware fault is detected by the in-service audit process. The in-service audits periodically run diagnostics on some critical EIU hardware components. The audits run every minute, each time checking one quarter of the hardware components.

In-service diagnostics also include an idler class audit, which runs continuously. The idler class runs whenever there are no other useful processes running. The idler class audit tests the EIC shared RAM. The idler class audit detects the missing Ethernet interface card (EIC) card almost instantly. The EIU goes SysB immediately when EIC is pulled out. The missing EIC causes EIC memory fault, which causes the idler process to TRAP.

In-service leaky bucket audit

Several LAN errors can be detected by the EIC diagnostics software. However, these errors are not severe or fatal enough to be reported on every occurrence of the error. Most of these errors occur when the Ethernet LAN is carrying above average traffic. These errors are transient and therefore do not require the EIU to take any action. However, if these errors occur frequently, relative to the Ethernet traffic, it may be indicative of a faulty component on the LAN. The leaky bucket algorithm is applied to such errors before reporting them through in-service trouble alarm and PM logs.

The leaky bucket algorithm measures errors against relative traffic and therefore allows the EIU to report faults only when a preset error count threshold is exceeded. The algorithm is controlled by five parameters:

•leaky bucket size

•fault detection threshold mark. The fault is reported when the bucket level exceeds this value.

•fault clear threshold mark. The fault reported previously, is cleared when the bucket level is lower than this value.

•error event factor. This value is added to the bucket level when an error event occurs.

•good event factor. This value is subtracted from the bucket level when a good event occurs.

DMS-100 Family EIU User Guide TELECOM12