Move metric descriptiions to package vars to avoid allocating them every
time `NewCPUFreqCollector()` is called.
Signed-off-by: Ben Kochie <superq@gmail.com>
This change updates the perf-utils library to 0.6.0 which has some fixes
for automatically detecting the correct tracefs mountpoint if available.
Signed-off-by: Daniel Hodges <hodges.daniel.scott@gmail.com>
With `--runtime.gomaxprocs=0`, the GOMAXPROXS value will default to the
number of logical CPUs. In this case, it is more useful to log the
actual value than the value set by the user via the command-line.
Signed-off-by: Simon Pasquier <spasquie@redhat.com>
The --web.config flag changed to --web.config.file in
440a132c38 and was realised in the recent
v1.5.0 release.
Signed-off-by: Joe Groocock <me@frebib.net>
NOTE: This changes the Go runtime "GOMAXPROCS" to 1. This is done to limit the
concurrency of the exporter to 1 CPU thread at a time in order to avoid a
race condition problem in the Linux kernel (#2500) and parallel IO issues
on nodes with high numbers of CPUs/CPU threads (#1880).
* [CHANGE] Default GOMAXPROCS to 1 #2530
* [FEATURE] Add multiple listeners and systemd socket listener activation #2393
* [ENHANCEMENT] Add RTNL version of netclass collector #2492, #2528
* [BUGFIX] Fix diskstats exclude flags #2487
* [BUGFIX] Bump go/x/crypt and go/x/net #2488
* [BUGFIX] Fix hwmon label sanitizer #2504
* [BUGFIX] Use native endianness when encoding InetDiagMsg #2508
* [BUGFIX] Fix btrfs device stats always being zero #2516
* [BUGFIX] Security: Update exporter-toolkit (CVE-2022-46146) #2531
Signed-off-by: Ben Kochie <superq@gmail.com>
Avoid running on all CPUs by limiting the Go runtime to one CPU by
default. Avoids having Go routines schedule on every CPU, driving up the
visible run queue length on high CPU count systems.
This also helps workaround a kernel deadlock issue with reading from
sysfs concurrently.
See:
* https://github.com/prometheus/node_exporter/issues/1880
* https://github.com/prometheus/node_exporter/issues/2500
Signed-off-by: Ben Kochie <superq@gmail.com>
* Refactor netclass_rtnl collector
Merge the netclass_rtnl collector into the netclass collector.
* Disabled by default
* Followup to #2492
Signed-off-by: Ben Kochie <superq@gmail.com>
* update rtnetlink package to v1.2.3
* add RTNL version of netclass collector that have all the metrics that netdev collector provides, too.
Signed-off-by: Haoyu Sun <hasun@redhat.com>
Copr community prometheus-exporters repository is obsoleted.
Signed-off-by: Otto Sabart <seberm@seberm.com>
Signed-off-by: Otto Sabart <seberm@seberm.com>
Some systems have broken netlink messages due to patched kernels. Since
these messages can not be parsed, add a flag to fall back to parsing
from `/proc/net/dev`.
Fixes: https://github.com/prometheus/node_exporter/issues/2502
Signed-off-by: Ben Kochie <superq@gmail.com>
Signed-off-by: Ben Kochie <superq@gmail.com>
Note however that the InetDiagMsg struct contains a InetDiagSockID
member, which itself contains some members which are explicitly
specified as big-endian in Linux kernel source:
struct inet_diag_sockid {
__be16 idiag_sport;
__be16 idiag_dport;
__be32 idiag_src[4];
__be32 idiag_dst[4];
__u32 idiag_if;
__u32 idiag_cookie[2];
};
node_exporter currently does not use these members for anything, so this
is acceptable (for now).
Signed-off-by: Daniel Swarbrick <daniel.swarbrick@gmail.com>
* docs/node-mixin: add fsMountpointSelector
This adds the option to add a `mountpoint` selector to filesystem
related alerts. The default is `mountpoint!=""`.
* docs/node-mixins: add fsMountpointSelector to dashboards
Signed-off-by: Jan Fajerski <jfajersk@redhat.com>
