prometheus/cadvisor
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity
2,184 Commits 1 Branch 39 Tags 50 MiB
cfb16f102b
Commit Graph

342 Commits

Author SHA1 Message Date
Euan Kemp
d2e11efba2 libcontainer: use real number of CPUs for usage 
As of the 4.7 kernel, the cpustats field returned from libcontainer
contains values for every possible cpu (including nonexistent ones).
The extra values are all 0s.

If we assume that hotplug events won't happen, we can get a more
accurage cpu count by using runtime.NumCPU and then ignoring any values
beyond that.
 2017-08-30 14:26:26 -07:00
Karl
5a033c064e Add runtime options for TLS support 2017-08-24 15:14:33 +02:00
Derek Carr
6fa48d9048 Expose total_rss when hierarchy is enabled 2017-08-23 14:56:59 -04:00
Derek Carr
d493f11f0b Reduce log spam when unable to get network stats 2017-08-18 16:11:03 -04:00
Andrew Pilloud
9e40f1c837 Fix race in ignoring non-existent files 2017-06-08 11:16:31 -07:00
David Ashpole
96b603e20a Merge branch 'master' into overlay2 2017-05-31 14:05:54 -07:00
David Ashpole
de308101e4 Merge branch 'master' into container-restart-count 2017-05-30 11:33:39 -07:00
Ricardo Pchevuzinske Katz
f5a896ef4c Gather information from RestartCount counter and put into labels 2017-05-30 14:05:14 -03:00
Andy Goldstein
69c085d9fa Cache the Docker thin pool name 
Cache the Docker thin pool name in the dockerFactory and pass it to each Docker container handler,
instead of calling `docker info` each time a new container handler is created, as the thin pool name
is extremely unlikely to change.
 2017-05-30 11:50:58 -04:00
Clayton Coleman
4e25a7951f
Report container FS metrics into prometheus /metrics 
PerDiskStats reported from cgroups were not being surfaced into
prometheus. In order to properly correlate the metrics, we need to
assign a device label to each metric (which is the FS or device path).
Since blkio cgroup tracks devices, we create a synthetic device
`/dev/NAME` for the metric.

Assign a Device label to each PerDiskStat for the handlers up front, and
then surface the PerDiskStat values into the prometheus metrics. Report
two new metrics - total bytes read and total bytes written.
 2017-04-24 20:46:55 -04:00
Bilal Amarni
e0c8ec569d [docker] add overlay2 storage driver 
Signed-off-by: Bilal Amarni <bilal.amarni@gmail.com>
 2017-04-13 13:39:45 +02:00
Tristan Colgate
227bb3a79d Add udp and udp6 network statistics 2017-04-10 20:41:51 +01:00
Manjunath A Kumatagi
8fb1158353 Add Docker API version 2017-04-04 10:56:11 +05:30
Brian Akins
feecd47daa Add watcher for zfs similar to devicemapper 
minor cleanup

ensure we look at parent dataset for limit, etc
 2017-03-15 18:31:11 -04:00
Derek Carr
b172ba65b4 disable thin_ls due to excessive iops 2017-02-09 10:29:58 -05:00
Calum Lacroix
6d67091c8b Checks for successful string split on env variable 2017-02-08 22:39:34 +00:00
Calum Lacroix
967e09a3a5 Checks environment variables for empty strings 
If an environment variable is an empty string, cadvisor panics and fails to start
 2017-02-08 18:44:56 +00:00
Derek Carr
4ad5c58f6a Ignore any .mount cgroup in docker handler 2017-01-16 11:02:37 -05:00
David Ashpole
9fdeefe3e4 Cadvisor now publishes per-container inode stats using the command 'find . -xdev printf '.'| wc -c' this is published in the v2 api using a new field 2016-10-12 13:16:21 -07:00
Tim St. Clair
afe67fec68
Cleanup comment style: // should be proceded by a space 2016-10-07 17:06:02 -07:00
derekwaynecarr
b84046f12c Look at all cgroup mounts 2016-09-22 15:34:59 -04:00
Chun Chen
844c98e073 GetSpec missing memory reservation 2016-08-16 23:26:46 +08:00
Florian Koch
3ce98a46c4 add cgropu swap usage and export as prometheus metric 2016-08-09 07:33:37 +02:00
Tobias Schmidt
1653733ea7 Expose cpu cgroup CFS prometheus metrics 
If CPU quota is configured (cpu.cfs_quota != -1) the CFS will provide
stats about elapsed periods and throtting in cpu.stats. This change
makes these information available as container_cpu_cfs_* metrics.
 2016-08-06 18:08:26 -04:00
Andy Goldstein
2b525ff87e Ensure minimum kernel version for thin_ls 
Ensure that kernel >= 4.4.0 or RHEL/Centos 7 kernel >= 3.10.0-366 exists before starting the thin
pool watcher. Prior versions have a bug in which reserving and releasing the metadata snapshot can
cause thin pool corruption.
 2016-08-04 16:14:40 -04:00
derekwaynecarr
6c114be580 Expose total inodes 2016-08-02 10:47:51 -04:00
Tim St. Clair
bbe54c0a10 Merge pull request #1368 from derekwaynecarr/inodes_pointer 
Allow clients to know if inodes data is available
 2016-07-29 10:10:30 -07:00
derekwaynecarr
cccf9d5fec Allow clients to know if inodes are supported on a filesystem 2016-07-26 11:15:07 -04:00
Andy Goldstein
d05098b314 Lower ThinPoolWatcher cache miss log level 2016-07-23 07:07:42 -04:00
Michael Taufen
307d1b1cb3 Modify working set memory stats calculation 
Change working set calculation to usage - total_inactive_file, rather than
usage - total_inactive_anon - total_inactive_file. Since writes to tmpfs
get tracked as total_inactive_anon when swap is disabled, the old
calculation would under-report memory pressure.

See this Kubernetes issue for context:
https://github.com/kubernetes/kubernetes/issues/28619
 2016-07-15 10:58:25 -07:00
Matt Wringe
213c254bb7 Add in the ability to expose the ip address of a container. 2016-07-13 13:58:43 -04:00
Tim St. Clair
35f7bc5ee7 Move mocks to testing package to remove +build tags 
Some go tools (e.g. godef, gorename) don't handle +build tags well, so
I refactored some packages to remove the test tags from cAdvisor.
 2016-07-06 14:15:43 -07:00
Seth Jennings
eafff745a5 only fail fs stat gather if metadata device not found 2016-06-21 16:05:37 -05:00
Paul Morie
c37d048e1b Check for thin_ls binary in path when using ThinPoolWatcher 2016-06-14 13:19:05 -04:00
Tim St. Clair
3fbe18de9a Merge pull request #1204 from pmorie/dm-support 
devicemapper thin_ls support
 2016-05-18 12:30:05 -07:00
Paul Morie
647224c95a Add devicemapper support for docker containers 2016-05-18 15:19:20 -04:00
Shaya Potter
6fa3687720 Polling rkt implementation of new watcher inteface (#1284) 
polling rkt implementation of new watcher inteface
 2016-05-17 10:34:56 -07:00
Shaya Potter
e02632463b Refactor container watching out of raw handler into its own inteface / package 2016-05-11 20:27:10 -07:00
Tim St. Clair
8404284c42 Merge pull request #1274 from timstclair/container 
Delete unused ListThreads container method
 2016-05-09 11:59:28 -07:00
Tim St. Clair
d1868287db Clean up docker root dir 2016-05-06 19:31:36 -07:00
Tim St. Clair
4c506006f2 Don't validate docker state file, since it's no longer used 2016-05-06 19:29:24 -07:00
Tim St. Clair
019bfaecc7 Delete unused ListThreads container method 2016-05-06 13:45:29 -07:00
Tim St. Clair
05fb225624 Merge pull request #1269 from sjpotter/update-rkt 
version checking for rkt + godep changes
 2016-05-06 12:40:40 -07:00
Shaya Potter
44795d7ce0 add version checking for rkt api service + godep changes 2016-05-06 01:01:20 -07:00
Vish Kannan
bebe18b67a Merge pull request #1261 from sjpotter/listToCommon 
move ListContainers code to common and make rkt play nice
 2016-05-03 10:12:54 -07:00
Shaya Potter
5ca11bb1a0 move ListContainers code to common and make rkt play nice 
playing nice means not elliding the system.slice cgroup but allowing the
raw handler to handle it
 2016-05-03 07:14:53 -07:00
Tim St. Clair
4d3ef349fb Move utils/machine -> machine 2016-05-02 15:56:49 -07:00
Tim St. Clair
f365c6a115 Move docker types to v1 API 2016-05-02 15:52:29 -07:00
Tim St. Clair
0c89fd1b71 Refactor docker-specific functions from manager to docker 2016-05-02 12:24:31 -07:00
Lantao Liu
ece4c555cc switch to the new engine-api 2016-04-25 19:22:05 -07:00