Return
-
Title
Storage: Brocade non-link errors are extremely high on the storage dashboard -
Description
Storage: Brocade non-link errors are extremely high on the storage dashboard -
Cause
Foglight reports the "Non-Link Error Rate" (per-port) from what it collects from the Brocade's "err_bad_os (Invalid Ordered Set)" metric (among others) and this can be quite high (10s of millions). The err_bad_os counter specifically is not a critical error and can be ignored.
The command portstatsshow <portnumber>, on the Brocade, will show the metric err_bad_os / InvalidOrderedSet.
Note: High values for the err_bad_os counter, while not serious in and of themselves, might be masking other other non-link errors on the port -
Resolution
To reduce the Non-link error rate run the following per port on the Brocade:
- execute "portshow <portnumber>"
- execute "portstatsshow <portnumber>"
- Select one of the ports reporting high non-link error rates
- execute "Portdisable <portnumber>"
- execute "PortCfgfillword <portnumber>, 3"
- execute "Portenable <portnumber>"
-
Additional Information
The following values from Brocade are added together to get the *non* link errors: • InvalidOrderedSets • RxTruncFrames • AddressErrors • DelimiterErrors • EncodingDisparityErrors • TooManyRdys • RxCRCs • RxBadEOFs • RxEncOutFrames • McastTimeouts • FRJTFrames • PRJTFrames • CRCErrors • InvalidTransmissionWords • PrimitiveSeqProtocolErrCount • ErrorFrames • FramesTooShort • FramesTooLong • Class1FRJT • Class1PRJT • Class2FRJT • Class2PRJT The following values Brocade are added together to get the link errors: • LinkResetsReceived • LinkResetsTransmitted • TxOfflineSequences • RxOfflineSequences • LinkFailures • LossOfSignalCounter • LossOfSyncCounter • NOSCount The names of these fields are from the SMI output of BNA, but it should be matched to the "portstatshow" fairly easily.