[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: prune images from registry - getsockopt: connection timed out





On Wed, Oct 11, 2017 at 10:44 AM, Maciej Zarczynski <m zarczynski adbglobal com> wrote:
Hi,

for a while we are running openshift origin standalone docker-registry
and it works pretty well, but when i try to prune images i am facing
following situation:

[root c37ee07bf04f /]# time oc adm prune images --keep-tag-revisions=10 --confirm

error: error communicating with registry 172.30.146.162:5000: [Get https://172.30.146.162:5000/healthz: dial tcp 172.30.146.162:5000: getsockopt: connection timed out, Get http://172.30.146.162:5000/healthz: dial tcp 172.30.146.162:5000: getsockopt: connection timed out]


it looks like the machine you're running prune from doesn't have access to the cluster network.  can you run the command from one of your cluster nodes?

 

real    4m38.046s

user    0m9.118s

sys    0m0.529s

[root c37ee07bf04f /]# time oc adm prune images --keep-tag-revisions=10000 --confirm

error: error communicating with registry 172.30.146.162:5000: [Get https://172.30.146.162:5000/healthz: dial tcp 172.30.146.162:5000: getsockopt: connection timed out, Get http://172.30.146.162:5000/healthz: dial tcp 172.30.146.162:5000: getsockopt: connection timed out]

real    4m37.320s

user    0m9.141s

sys    0m0.564s

[root c37ee07bf04f /]# time oc adm prune images --keep-tag-revisions=10 | wc

Dry run enabled - no modifications will be made. Add --confirm to remove images

  27771   53320 2861369

real    0m23.583s

user    0m10.540s

sys    0m0.576s

[root c37ee07bf04f /]# time oc adm prune images --keep-tag-revisions=10000 | wc

Dry run enabled - no modifications will be made. Add --confirm to remove images

     47      53    3031

real    0m23.728s

user    0m9.060s

sys    0m0.465s

[root c37ee07bf04f /]# oc get svc -n default

NAME               CLUSTER-IP       EXTERNAL-IP   PORT(S)                   AGE

docker-registry    172.30.146.162   <none>        5000/TCP                  138d

kubernetes         172.30.0.1       <none>        443/TCP,53/UDP,53/TCP     138d

registry-console   172.30.142.214   <none>        9000/TCP                  138d

router             172.30.105.200   <none>        80/TCP,443/TCP,1936/TCP   138d

[root c37ee07bf04f /]# oc version

oc v3.6.0+c4dd4cf

kubernetes v1.6.1+5115d708d7

features: Basic-Auth GSSAPI Kerberos SPNEGO

Server https://intentionally.removed.com:8443

openshift v1.5.0+031cbe4

kubernetes v1.5.2+43a9be4


As you can see above, dry run works without problem and return a bunch
of elements for removeal but when flag --confirm is added problem appears.

At fist i thought that docker-registry pod i failing with healthchecks
but after after some investigation it is probably not the cause, at
least i don't see any events for docker-registry pod.

Is it possible to change some timeouts values for origin-master ? (I
suspect that the error is caused by origin-master which breaks
connection after ~ 277s from oc cli to registry)

I was also thinking about workaround: Use output from dry-run and pass
to some other tool (skopeo maybe?).

Have you ever faced such problem?


Best Regards,
Maciej Żarczyński


[https://www.adbglobal.com/wp-content/uploads/adb.png]
adbglobal.com<https://www.adbglobal.com>
[https://www.adbglobal.com/wp-content/uploads/linkedin_logo.png]<https://www.linkedin.com/company/adb/>         [https://www.adbglobal.com/wp-content/uploads/twitter_logo.png] <https://twitter.com/adb_global>        [https://www.adbglobal.com/wp-content/uploads/pinterest_logo.png] <https://pinterest.com/adbglobal/pins/>

_______________________________________________
users mailing list
users lists openshift redhat com
http://lists.openshift.redhat.com/openshiftmm/listinfo/users



--
Ben Parees | OpenShift


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]