Bitbucket pipeline times out sending command to Kubernetes

[Aias Martakis]

Hello all,

I've been using Bitbucket pipelines successful for my project for a while and it's great!

Two days ago, however, I started receiving a time-out error on (only) the last command (where k8s pulls the image from the docker hub repository).

+ kubectl set image deployment/ingestionclient ingclient=someuser/ingclient:$BITBUCKET_COMMITUnable to connect to the server: dial tcp 34.199.150.69:443: i/o timeout

Initially I was convinced it had to be a networking issue as I would assume the above statement simply instructs the k8s cluster to pull the container images from docker hub.

Interestingly if i perform the above command from my local laptop (same cluster, same image) it executes with no problem! On top of that i received feedback today from my networking department confirming no changes were made to the network.

Any idea if this can be related to anything in bitbucket pipelines? Note that we added significant amount of code in the last few weeks and so I've been forced to set memory parameters, giving 3072mb to the docker service (leaving 1024 for the rest) in pipelines so it compiles the javascript application in the dockerfile (but this hasn't been an issue before either).

Another thing to note is that we run an old 1.8.7 version of k8s that I know is incompatible with the latest kubectl, so i force pipelines to pull the 1.9.10 version which has worked before.

Any help/hints where to look further would be much appreciated.

Thanks,

Aias

[Aias Martakis]

I can't believe it but as soon as I was done writing this, we recompiled and reran the pipeline file without making any changes...and guess what it now works again!

Must've been some network glitch? or one with docker hub?