failed to persist metadata - open nsqd.308.dat.xxx.tmp: permission denied

[Cryptophobia]

From @amarburg on October 10, 2016 19:48

Hi all, just installed Workflow onto a 3-node Vagrant test network using your great instructions. Unfortunately, nsqd goes into an error-restart:

% kubectl --namespace=deis log deis-nsqd-2362409882-od1bd
Set broadcast-address to and MAX_READY_COUNT to 10000
[nsqd] 2016/10/10 19:41:21.340182 nsqd v0.3.8 (built w/go1.6.2)
[nsqd] 2016/10/10 19:41:21.342169 ID: 308
[nsqd] 2016/10/10 19:41:21.342204 NSQ: persisting topic/channel metadata to nsqd.308.dat
2016/10/10 19:41:21 ERROR: failed to persist metadata - open nsqd.308.dat.2082426343634159740.tmp: permission denied

Not even sure where to start with this one -- is nsqd failing for some other reason then failing to persist the metadata, or is that the root problem.

Kube 1.3.8 on a hand-rolled CoreOS/Vagrant configuration, so it's entirely possible I've screwed something up. However, other cursory Kubernetes tests appear to work.

Copied from original issue: deis/nsq#9

[Cryptophobia]

From @bacongobbler on October 31, 2016 16:25

Hey @amarburg, sorry that this issue fell through the cracks!

I see the following upstream issue but I'm not sure how this issue may occur. Perhaps @jchauncey knows more on how this issue pops up.

[Cryptophobia]

From @amarburg on November 1, 2016 16:12

Hi @bacongobbler, thanks! Pls let me know how I can assist with the debugging. I'm a Kube newbie ...

[Cryptophobia]

From @bacongobbler on November 1, 2016 16:46

If you're up for it you can take a look at https://deis.com/docs/workflow/troubleshooting/kubectl/ to dive into the nsq component, take a look at their issue list and see if anything looks relevant. That's where I'd start, at least. :)