Part of the difficulty is that we wrap the net-snmp C library in Go. The error response of the library isn't all that easy to work with.
There are some options for pure-Go MIB parsing now, but converting the generator to this would take some work. On Sat, Jul 31, 2021 at 5:05 PM Brian Candler <[email protected]> wrote: > If you run the generator, and it can't find a required MIB, it stops to > give you an opportunity to install the correct MIB. But if I understand > you correctly, you're saying it's inconvenient to install one missing MIB, > run generator again, only to find there's another missing MIB, and so on. > You'd rather see all the problems at once. > > In that case, I think a better solution would be to continue running, and > then abort at the end. No extra flag like "fail-on-oid-notfound" would be > required then. > > However I suspect it could generate a very noisy output, because MIBs have > dependencies on each other: one missing MIB high up in the tree could > result in many errors being reported. Equally, many tables may all come > from the same missing MIB, so listing all the table names may not be all > that helpful. > > However, you could certainly try implementing this and see if it behaves > sensibly. Just print an error and set a fail flag on oidNotFound. > > On Saturday, 31 July 2021 at 14:45:25 UTC+1 [email protected] wrote: > >> Yes, this is a proposal. >> >> Others have told me that I need to monitor some OIDs and have provided a >> private MIB file, but when the generator works, it only tells me the first >> OID not found. >> >> In this case, I need to delete the added OID and run again, repeating >> several times to find all the invalid OIDs. >> >> >> >> 在2021年7月31日星期六 UTC+8 下午4:52:25<Brian Candler> 写道： >> >>> I'm sorry, it's not clear: is this a question? A statement? Or a >>> proposal for something to change? >>> >>> My best guess is it's the latter. If so, can you provide a use case >>> when this would be helpful? >>> >>> I note that this is a change to the generator, not the exporter. It >>> means that if you run the generator, but have forgotten to install the >>> necessary MIBs, then it will carry on and ignore the missing MIBs, and >>> therefore generate an incorrect/incomplete snmp.yml output. >>> >>> Maybe you don't have those MIBs, but surely in this case it would be >>> better to update generator.yml to remove the MIBs you are not interested >>> in? This makes it explicit that you don't want these MIBs to be included >>> in snmp.yml. >>> >>> On Saturday, 31 July 2021 at 07:02:47 UTC+1 [email protected] wrote: >>> >>>> The generate subcommand adds an option "fail-on-oid-notfound" default >>>> is "true". When "false", we print the "warn" or" error" log instead of >>>> exiting the program so that we can see all the missing oids. >>>> >>>> example possible: >>>> >>>> >>>> // Find all top-level nodes. >>>> >>>> metricNodes := map[*Node]struct{}{} >>>> >>>> for _, oid := range toWalk { >>>> >>>> metricNode, oidType := getMetricNode(oid, node, nameToNode) >>>> >>>> switch oidType { >>>> >>>> case oidNotFound: >>>> >>>> >>>> if *failOnOidNotFound { >>>> return nil, fmt.Errorf("cannot find oid '%s' to walk", oid) >>>> } >>>> level.Warn(logger).Log("msg", " cannot find oid", "oid", oid) >>>> continue >>>> >>>> case oidSubtree: >>>> >>>> needToWalk[oid] = struct{}{} >>>> >>>> case oidInstance: >>>> >>>> // Add a trailing period to the OID to indicate a "Get" instead of a >>>> "Walk". >>>> >>>> needToWalk[oid+"."] = struct{}{} >>>> >>>> // Save instance index for lookup. >>>> >>>> index := strings.Replace(oid, metricNode.Oid, "", 1) >>>> >>>> >>>> tableInstances[metricNode.Oid] = append(tableInstances[metricNode.Oid], >>>> index) >>>> >>>> case oidScalar: >>>> >>>> // Scalar OIDs must be accessed using index 0. >>>> >>>> needToWalk[oid+".0."] = struct{}{} >>>> >>>> } >>>> >>>> metricNodes[metricNode] = struct{}{} >>>> >>>> } >>>> >>>> -- > You received this message because you are subscribed to the Google Groups > "Prometheus Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > https://groups.google.com/d/msgid/prometheus-users/fd5b01d0-426d-4a63-a0ee-1013866c9ea0n%40googlegroups.com > <https://groups.google.com/d/msgid/prometheus-users/fd5b01d0-426d-4a63-a0ee-1013866c9ea0n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- You received this message because you are subscribed to the Google Groups "Prometheus Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/CABbyFmpvO_YFQStLuGs%2Bs__T_hrXkhWBL7FeaN567dVM78uXhg%40mail.gmail.com.
