-
Notifications
You must be signed in to change notification settings - Fork 1.1k
SHM size must be greater then 0 #2147
Comments
@EamonZhang
:) |
Swarm 1.2.0 Sorry, I do not know how to debug . docker -H :4000 --debug=true run --name nginx22 nginx |
@EamonZhang First, you run your swarm in container or just binary on host?
Second, when you can reproduce this issue, check the log swarm output and attach the log here Actually, you specified the versions. And I can try to reproduce this. :) I have finished my testing, while can not reproduce this:
|
@allencloud Thank you . 10.2.0.13 node3 manage below command on node3
|
I've seen this as well, but setting --shm-size=64m on build and run makes it work again. 64m is the default. |
I should add, I'm also using CentOS7. I'm using the swarm:latest container. We're doing master replication and it only seems to happen if the call is made from the same node. If I specify --shm-size, it runs fine, but this also breaks running a container with Ansible or Jenkins. I haven't done regression testing yet to find out when it was broken release-wise, but we started seeing it about 2-3 weeks ago. |
I've found the issue does not occur with swarm 1.1.3, but does with 1.2.0 and 1.2.1-rc1. |
I think it's #1879. engine-api needs shm-size. It was merged in docker issue 16168. Docker engine version needs at least 1.10. |
Have the same problem with:
|
RUN docker swarm :
docker -H :4000 run --name nginx22 nginx
Error response from daemon: 500 Internal Server Error: SHM size must be greater then 0
Change command
docker -H :4000 run --name nginx22 --shm-size=67108864 nginx
ok!
docker code :
const DefaultSHMSize int64 = 67108864
Is there anything overwrite DefaultSHMSize = 0
CentOS 7
df -h
tmpfs 497M 0 497M 0% /dev/shm
The text was updated successfully, but these errors were encountered: