The following 5 patches extend the functionality of ibqueryerrors to cover some of the functionality of the ib* bash scripts.
ibqueryerrors has been our workhorse for fabric monitoring here at LLNL. While I am working on better performance management in OpenSM this provides a scalable "direct" option for querying the fabric. The main thrust for replicating the shell scripts is the fact that ibqueryerrors is an order of magnitude faster than the corresponding shell scripts. For example, on a "small" 141 node cluster. Gathering the error counters takes 22 seconds vs < 1 second with ibqueryerrors [*] hype137@root:time ./ibcheckerrors > /dev/null real 0m22.797s user 0m3.284s sys 0m9.783s hype137@root:time ./ibqueryerrors > /dev/null real 0m0.348s user 0m0.051s sys 0m0.113s Gathering all the counter data takes 28 seconds vs < 2 with ibqueryerrors [*] hype137@root:time ./ibdatacounters > /dev/null real 0m28.469s user 0m4.128s sys 0m11.560s hype137@root:time ./ibqueryerrors --counters > /dev/null real 0m1.262s user 0m0.410s sys 0m0.643s In addition, PortCountersExtended are now supported. Ira [*] Times shown are representative of single runs but are on the same order of my testing over many runs. Other factors such as the number of outstanding SMP's, system load, or possible fabric issues, may change performance. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
