ksync not working at all on an EKS cluster; syncthing process stops running

[aberle]

Creating a simple spec to watch a directory with a single file starts to run, but then syncthing stops running.

mkdir ksync_test
touch ksync_test/a-file
ksync -n my-namespace create --selector=name=internal-model-service /home/aberle/111/ksync_test /var/www/111/ksync_test

After a while, kysnc watch will just start logging messages that look like:

DEBU[0019] RESTY 2021/01/27 11:29:44 ERROR Get "http://localhost:8384/rest/events?since=30": dial tcp 127.0.0.1:8384: connect: connection refused, Attempt 2 

and nothing works until I restart. I see a defunct syncthing process running when this happens:

$ ps aux | grep syncthing                                                                                                                                                                           
aberle    3692  0.0  0.0      0     0 pts/7    Z+   11:29   0:00 [syncthing] <defunct>

All the matching pods are found, but their STATUS is stuck at starting:

       NAME          LOCAL              REMOTE             STATUS                POD                CONTAINER  
-----------------+------------+-------------------------+----------+------------------------------+------------
  upward-dolphin   ksync_test   /var/www/111/ksync_test             
                                                          starting   internal-model-service-9mqf9              
                                                          starting   internal-model-service-29ts4              
                                                          starting   internal-model-service-8dfxj              
                                                          starting   internal-model-service-lr78h 

If I run ksync watch a few times after a crash it will sometimes work, but then stop working on subsequent restarts of ksync watch. When it "works" I still see syncthing restart several times.

This is running on EKS, K8S version 1.16:

$ kubectl version   
                                                                                                                                                                                
Client Version: version.Info{Major:"1", Minor:"16", GitVersion:"v1.16.13", GitCommit:"39a145ca3413079bcb9c80846488786fed5fe1cb", GitTreeState:"clean", BuildDate:"2020-07-15T16:18:19Z", GoVersion:"go1.13.9", Compiler:"gc", Platform:"linux/amd64"}
Server Version: version.Info{Major:"1", Minor:"16+", GitVersion:"v1.16.15-eks-ad4801", GitCommit:"ad4801fd44fe0f125c8d13f1b1d4827e8884476d", GitTreeState:"clean", BuildDate:"2020-10-20T23:27:12Z", GoVersion:"go1.13.15", Compiler:"gc", Platform:"linux/amd64"}

Here are the logs from ksync watch --log-level debug from one instance of this happening: ksync_watch.log

ksync is very well suited to my development process, so I'd really like to get this working, thanks.

[ntorba]

@aberle Did you ever find a fix for this? I'm seeing the same issue on my EKS cluster

[aberle]

@aberle Did you ever find a fix for this? I'm seeing the same issue on my EKS cluster

Sorry, no. I gave up on using ksync since this was not working for me.

[timfallmk]

The error messages you're seeing are from the local ksync container being unable to connect to the (also local) syncthing instance. Probably a local networking problem (if you running containers in a vm locally, the networking setup here can cause this).