[PVE-User] invoking ssh crashes container

[Fabian Grünbichler]

On June 28, 2021 5:09 pm, Adam Weremczuk wrote:
> Hi all,
> 
> I run a minimalistic Debian 10.10 container on Proxmox 6.2-6 dual host 
> stack.
> 
> It has 512 MB of memory assigned and most of the time uses about 10%.
> 
> Every time I try to ssh from it (using bash in web console) to a 
> different host the container crashes (powers off).
> 
> Syslog records the following:
> 
> (...)
> Jun 28 13:51:46 debian10 kernel: [26284286.210180] [  25629]  1006 
> 25629     3205     1290    61440        0             0 ssh
> Jun 28 13:51:46 debian10 kernel: [26284286.210188] [  25634]  1006 
> 25634     3205     1289    61440        0             0 ssh
> Jun 28 13:51:46 debian10 kernel: [26284286.210196] [  25640]  1006 
> 25640     3205     1292    65536        0             0 ssh
> Jun 28 13:51:46 debian10 kernel: [26284286.210207] [  23745]     0 
> 23745     4880     1753    73728        0             0 systemd-logind
> Jun 28 13:51:46 debian10 kernel: [26284286.210218] [    810] 0   810     
> 3962     1576    65536        0         -1000 sshd
> Jun 28 13:51:46 debian10 kernel: [26284286.213678] ssh invoked 
> oom-killer: gfp_mask=0x100cca(GFP_HIGHUSER_MOVABLE), order=0, 
> oom_score_adj=0
> Jun 28 13:51:46 debian10 kernel: [26284286.213729] dump_header+0x4f/0x1e1
> Jun 28 13:51:46 debian10 kernel: [26284286.213770] 
> mem_cgroup_try_charge+0x71/0x190
> Jun 28 13:51:46 debian10 kernel: [26284286.213820] filemap_fault+0x86f/0xa60
> Jun 28 13:51:46 debian10 kernel: [26284286.213971] 
> __handle_mm_fault+0xeb0/0x12e0
> Jun 28 13:51:46 debian10 kernel: [26284286.214035] RIP: 0033:0x5619b5c48930
> Jun 28 13:51:46 debian10 kernel: [26284286.214286] [  24101]  1006 
> 24101     3205     1088    65536      172             0 ssh
> Jun 28 13:51:46 debian10 kernel: [26284286.214293] [  24105]  1006 
> 24105     3205     1108    65536      173             0 ssh
> Jun 28 13:51:46 debian10 kernel: [26284286.214299] [  24109]  1006 
> 24109     3205     1103    65536      173             0 ssh
> (...)
> MANUAL POWER ON
> Jun 28 13:58:49 debian10 systemd-sysctl[44]: Couldn't write '1' to 
> 'fs/protected_hardlinks', ignoring: Read-only file system
> Jun 28 13:58:49 debian10 systemd-sysctl[44]: Couldn't write '1' to 
> 'fs/protected_symlinks', ignoring: Read-only file system
> Jun 28 13:58:49 debian10 systemd[1]: Starting Flush Journal to 
> Persistent Storage...
> Jun 28 13:58:49 debian10 systemd[1]: Started Create System Users.
> (...)
> 
> Any ideas?

the log says it got OOM killed. did you enable persistent journal inside 
the container? if not, that might be the culprit (tmpfs usage in the 
container gets accounted towards the container's resource limits).

`mkdir /var/log/journal && systemctl restart systemd-journald`