deis-monitor-telegraf shows - error getting disk usage info: too many levels of symbolic links

[rimusz]

deis-monitor-telegraf-rz92o deis-monitor-telegraf 2016/12/02 16:03:30 E! ERROR in input [inputs.disk]: error getting disk usage info: too many levels of symbolic links

[bacongobbler]

Can you please provide some basic information on how you are seeing this:

• What version of workflow?
• What version of Kubernetes?
• How is it deployed?
• Is there a set of instructions to re-create this issue?
• Have you gone through the troubleshooting guide?

[rimusz]

• latest Workflow v2.8.0 deployed with Helm chart
• k8s v1.4.6
• deployed on AWS, CoreOS, similar to tack setup @gerred can explain more on the setup
• no, there are no set of instructions to re-create this issue
• yes, only error I got is deis-monitor-telegraf-rz92o deis-monitor-telegraf 2016/12/02 16:03:30 E! ERROR in input [inputs.disk]: error getting disk usage info: too many levels of symbolic links

[wetoolaguer]

Hi, I experienced the same issue.

• Workflow versions v2.9.0, v2.5.0 with helm and helmc
• deployed on AWS, debian / ubuntu via kops
• I get 2016/12/14 11:43:10 ERROR in input [disk]: error getting disk usage info: too many levels of symbolic links and one of my telegrafs intermittently fails with this log

Creating topic with URL: http://100.71.131.206:4151/topic/create?topic=metrics

[bacongobbler]

relevant: influxdata/telegraf#1352

[bacongobbler]

This does appear to be an issue upstream with systemd, perhaps with https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/1555760.

[mboersma]

There are some workarounds suggested in influxdata/telegraf#1352 that users can try. A pervasive fix requires Ubuntu to make changes to systemd or binfmt, which is something many people are still waiting for sadly.

I'm going to move this to the next milestone since there's nothing currently actionable for Workflow here IMHO.

[monaka]

@vdice Was this issue fixed on v2.12 ? Or re-tagged on v2.13 or later?

[bacongobbler]

This issue was never resolved; see above

[monaka]

Ah, I see.

[michalkm]

Any updates or workarounds on this matter ?

[jchauncey]

See earlier in the thread. As it stands this is not a fix that we can make on the workflow end it has to be fixed in an upstream project.