Bug 1479854

Summary: Can't start more than 24 containers with systemd-233 inside
Product: [Fedora] Fedora Reporter: Viktor Ashirov <vashirov>
Component: systemdAssignee: systemd-maint
Status: CLOSED NOTABUG QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 26CC: johannbg, lnykryn, msekleta, muadda, ssahani, s, systemd-maint, zbyszek
Target Milestone: ---Keywords: Regression
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2017-08-30 15:56:50 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
Dockerfile and shell scripts none

Description Viktor Ashirov 2017-08-09 14:59:14 UTC
Created attachment 1311253 [details]
Dockerfile and shell scripts

Description of problem:
I'm running docker containers with Fedora 26 and systemd inside. But I can't start more than 24 (sometimes 25 or 26) containers. It's reproducible on different systems. Problem appeared in Fedora 26 with systemd-233. If I downgrade systemd in the container 232, I no longer observe the problem.

Version-Release number of selected component (if applicable):
systemd-233-6.fc26.x86_64

How reproducible:
always

Steps to Reproduce:
From the attached reproducer:
1. Build the container:
./build.sh 

2. Spin up 50 containers:
./run.sh

3. docker ps -q | wc -l
24

4. Optionally run cleanup.sh

Comment 1 Viktor Ashirov 2017-08-09 15:01:48 UTC
In the container that fails to start systemd, some mount points are absent, that present in the other containers:

configfs on /sys/kernel/config type configfs (rw,relatime)
debugfs on /sys/kernel/debug type debugfs (rw,relatime,seclabel)
systemd-1 on /proc/sys/fs/binfmt_misc type autofs (rw,relatime,fd=30,pgrp=1,timeout=0,minproto=5,maxproto=5,direct,pipe_ino=604626)

Comment 2 Viktor Ashirov 2017-08-30 15:56:50 UTC
After applying configuration for sysctl and limits.conf from https://github.com/lxc/lxd/blob/master/doc/production-setup.md problem has disappeared. Closing as NOTABUG.