diff --git a/CHANGELOG.next.asciidoc b/CHANGELOG.next.asciidoc index 7334d11e743..ceb3de1b608 100644 --- a/CHANGELOG.next.asciidoc +++ b/CHANGELOG.next.asciidoc @@ -70,6 +70,7 @@ https://github.com/elastic/beats/compare/v8.8.1\...main[Check the HEAD diff] - Fix handling of Juniper SRX structured data when there is no leading junos element. {issue}36270[36270] {pull}36308[36308] - Fix Filebeat Cisco module with missing escape character {issue}36325[36325] {pull}36326[36326] - Added a fix for Crowdstrike pipeline handling process arrays {pull}36496[36496] +- Fix TCP/UDP metric queue length parsing base. {pull}37714[37714] *Heartbeat* @@ -135,6 +136,7 @@ https://github.com/elastic/beats/compare/v8.8.1\...main[Check the HEAD diff] - Added support for Okta OAuth2 provider in the CEL input. {issue}36336[36336] {pull}36521[36521] - Added support for new features & removed partial save mechanism in the Azure Blob Storage input. {issue}35126[35126] {pull}36690[36690] - Added support for new features and removed partial save mechanism in the GCS input. {issue}35847[35847] {pull}36713[36713] +- Relax TCP/UDP metric polling expectations to improve metric collection. {pull}37714[37714] *Auditbeat* diff --git a/filebeat/input/tcp/input.go b/filebeat/input/tcp/input.go index 762c6b6ba35..1b3ffa7c2aa 100644 --- a/filebeat/input/tcp/input.go +++ b/filebeat/input/tcp/input.go @@ -238,31 +238,50 @@ func (m *inputMetrics) poll(addr, addr6 []string, each time.Duration, log *logp. // base level for the rx_queue values and ensures that if the // constructed address values are malformed we panic early // within the period of system testing. + want4 := true rx, err := procNetTCP("/proc/net/tcp", addr, hasUnspecified, addrIsUnspecified) if err != nil { - log.Warnf("failed to get initial tcp stats from /proc: %v", err) + want4 = false + log.Infof("did not get initial tcp stats from /proc: %v", err) } + want6 := true rx6, err := procNetTCP("/proc/net/tcp6", addr6, hasUnspecified6, addrIsUnspecified6) if err != nil { - log.Warnf("failed to get initial tcp6 stats from /proc: %v", err) + want6 = false + log.Infof("did not get initial tcp6 stats from /proc: %v", err) + } + if !want4 && !want6 { + log.Warnf("failed to get initial tcp or tcp6 stats from /proc: %v", err) + } else { + m.rxQueue.Set(uint64(rx + rx6)) } - m.rxQueue.Set(uint64(rx + rx6)) t := time.NewTicker(each) for { select { case <-t.C: + var found bool rx, err := procNetTCP("/proc/net/tcp", addr, hasUnspecified, addrIsUnspecified) if err != nil { - log.Warnf("failed to get tcp stats from /proc: %v", err) - continue + if want4 { + log.Warnf("failed to get tcp stats from /proc: %v", err) + } + } else { + found = true + want4 = true } rx6, err := procNetTCP("/proc/net/tcp6", addr6, hasUnspecified6, addrIsUnspecified6) if err != nil { - log.Warnf("failed to get tcp6 stats from /proc: %v", err) - continue + if want6 { + log.Warnf("failed to get tcp6 stats from /proc: %v", err) + } + } else { + found = true + want6 = true + } + if found { + m.rxQueue.Set(uint64(rx + rx6)) } - m.rxQueue.Set(uint64(rx + rx6)) case <-m.done: t.Stop() return @@ -323,10 +342,10 @@ func procNetTCP(path string, addr []string, hasUnspecified bool, addrIsUnspecifi } found = true - // queue lengths are decimal, e.g.: + // queue lengths are hex, e.g.: // - https://elixir.bootlin.com/linux/v6.2.11/source/net/ipv4/tcp_ipv4.c#L2643 // - https://elixir.bootlin.com/linux/v6.2.11/source/net/ipv6/tcp_ipv6.c#L1987 - v, err := strconv.ParseInt(string(r), 10, 64) + v, err := strconv.ParseInt(string(r), 16, 64) if err != nil { return 0, fmt.Errorf("failed to parse rx_queue: %w", err) } diff --git a/filebeat/input/udp/input.go b/filebeat/input/udp/input.go index 831fb41c2ee..cd7ca0c5605 100644 --- a/filebeat/input/udp/input.go +++ b/filebeat/input/udp/input.go @@ -231,33 +231,52 @@ func (m *inputMetrics) poll(addr, addr6 []string, each time.Duration, log *logp. // base level for the rx_queue and drops values and ensures that // if the constructed address values are malformed we panic early // within the period of system testing. + want4 := true rx, drops, err := procNetUDP("/proc/net/udp", addr, hasUnspecified, addrIsUnspecified) if err != nil { - log.Warnf("failed to get initial udp stats from /proc: %v", err) + want4 = false + log.Infof("did not get initial udp stats from /proc: %v", err) } + want6 := true rx6, drops6, err := procNetUDP("/proc/net/udp6", addr6, hasUnspecified6, addrIsUnspecified6) if err != nil { - log.Warnf("failed to get initial udp6 stats from /proc: %v", err) + want6 = false + log.Infof("did not get initial udp6 stats from /proc: %v", err) + } + if !want4 && !want6 { + log.Warnf("failed to get initial udp or udp6 stats from /proc: %v", err) + } else { + m.rxQueue.Set(uint64(rx + rx6)) + m.drops.Set(uint64(drops + drops6)) } - m.rxQueue.Set(uint64(rx + rx6)) - m.drops.Set(uint64(drops + drops6)) t := time.NewTicker(each) for { select { case <-t.C: + var found bool rx, drops, err := procNetUDP("/proc/net/udp", addr, hasUnspecified, addrIsUnspecified) if err != nil { - log.Warnf("failed to get udp stats from /proc: %v", err) - continue + if want4 { + log.Warnf("failed to get udp stats from /proc: %v", err) + } + } else { + found = true + want4 = true } rx6, drops6, err := procNetUDP("/proc/net/udp6", addr6, hasUnspecified6, addrIsUnspecified6) if err != nil { - log.Warnf("failed to get udp6 stats from /proc: %v", err) - continue + if want6 { + log.Warnf("failed to get udp6 stats from /proc: %v", err) + } + } else { + found = true + want6 = true + } + if found { + m.rxQueue.Set(uint64(rx + rx6)) + m.drops.Set(uint64(drops + drops6)) } - m.rxQueue.Set(uint64(rx + rx6)) - m.drops.Set(uint64(drops + drops6)) case <-m.done: t.Stop() return @@ -321,10 +340,10 @@ func procNetUDP(path string, addr []string, hasUnspecified bool, addrIsUnspecifi } found = true - // queue lengths and drops are decimal, e.g.: + // queue lengths and drops are hex, e.g.: // - https://elixir.bootlin.com/linux/v6.2.11/source/net/ipv4/udp.c#L3110 // - https://elixir.bootlin.com/linux/v6.2.11/source/net/ipv6/datagram.c#L1048 - v, err := strconv.ParseInt(string(r), 10, 64) + v, err := strconv.ParseInt(string(r), 16, 64) if err != nil { return 0, 0, fmt.Errorf("failed to parse rx_queue: %w", err) }