lockups since version 2:4.7.6+dfsg~ubuntu-0ubuntu2.9

[Valentijn Sessink]

Since version 2:4.7.6+dfsg~ubuntu-0ubuntu2.9 / Ubuntu 18.04.1 LTS, smbd locks up sometimes. We are unsure why. No messages in the log.smbd.

Symptoms: clients (a few, about 5 to 10, OSX and Linux machines) sometimes just cannot connect to Samba anymore. They call us, "the server doesn't work".

Log messages: none that we are alarmed by (smb.conf log level = 1)

Then, trying to restart samba (systemctl restart smbd.service) says:

Apr 12 14:23:37 kantoor systemd[1]: Stopping Samba SMB Daemon...
[...]
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: State 'stop-final-sigterm' timed out. Killing.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29016 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29017 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29034 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29035 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29046 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29047 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29048 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29049 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29067 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29068 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29077 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29078 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29487 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29749 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29784 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Killing process 29793 (smbd) with signal SIGKILL.
Apr 12 14:25:07 kantoor systemd[1]: smbd.service: Failed with result 'timeout'.
Apr 12 14:25:07 kantoor systemd[1]: Stopped Samba SMB Daemon.
Apr 12 14:25:07 kantoor systemd[1]: Starting Samba SMB Daemon...

While a regular restart, when there are no complaints about "the server isn't working", just goes in sub-second time, like this morning:
Apr 13 10:09:17 kantoor systemd[1]: Stopping Samba SMB Daemon...
Apr 13 10:09:17 kantoor systemd[1]: Stopped Samba SMB Daemon.
Apr 13 10:09:17 kantoor systemd[1]: Starting Samba SMB Daemon...
Apr 13 10:09:17 kantoor systemd[1]: Started Samba SMB Daemon.
Apr 13 10:09:17 kantoor smbd[14174]: [2019/04/13 10:09:17.107277, 0] ../lib/util/become_daemon.c:124(daemon_ready)
Apr 13 10:09:17 kantoor smbd[14174]: STATUS=daemon 'smbd' finished starting up and ready to serve connections

Would it help to try to attach "strace -p ... -p ..." for the smbd processes next time the lockup happens? Or do we need another log level?

The smb.conf is fairly straightforward (maybe a bit oldfashioned):
[global]
        log level = 1
        workgroup = hssa
        passdb backend = ldapsam:ldap://127.0.0.1/
        ldap admin dn = cn=admin,dc=kantoor,dc=hssa,dc=nl
        ldap ssl = off
        ldap suffix = dc=kantoor,dc=hssa,dc=nl
        ldap user suffix = ou=Users
        ldap group suffix = ou=Groups
        ldap machine suffix = ou=Computers
        unix extensions = yes
        delete readonly = yes
        ea support = yes
        ldap password sync = yes
        interfaces = 127.0.0.0/8 ens3
        bind interfaces only = true
        load printers = no
        printing = bsd
        printcap name = /dev/null
        disable spoolss = Yes
        disable netbios = yes
        smb ports = 445
        dns proxy = no
        vfs objects = fruit streams_xattr
        security = user

[homes]
        comment = Home Directories
        writable = yes

[hssa]
        force group = users
        force directory mode = 2770
        force create mode = 0660
        directory mask = 2770
        create mode = 0660
        comment = Algemeen
        writable = yes
        path = /home/hssa
        mangled names = no
        mangling char = _
        valid users = @users

[hssa-intern]
        force group = hssaintern
        force directory mode = 2770
        force create mode = 0660
        directory mask = 2770
        create mode = 0660
        comment = HSSA-intern
        writable = yes
        path = /home/hssa intern
        mangled names = yes
        mangling char = _
        valid users = @users
---
ApportVersion: 2.20.9-0ubuntu7.1
Architecture: amd64
DistroRelease: Ubuntu 18.04
NmbdLog:

OtherFailedConnect: Yes
Package: samba 2:4.7.6+dfsg~ubuntu-0ubuntu2.9
PackageArchitecture: amd64
ProcVersionSignature: Ubuntu 4.15.0-36.39-generic 4.15.18
SambaServerRegression: Yes
SmbConfIncluded: No
Tags: bionic
TestparmExitCode: 0
Uname: Linux 4.15.0-36-generic x86_64
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups:

_MarkForUpload: True

[Valentijn Sessink (valentijn)]

apport information

[Valentijn Sessink (valentijn)]

apport information

[Valentijn Sessink (valentijn)]

apport information

[Valentijn Sessink (valentijn)]

apport information

[Valentijn Sessink (valentijn)]

apport information

[Marc Deslauriers (mdeslaur)]

Thanks for reporting this issue. Could you try downgrading your packages to version 2:4.7.6+dfsg~ubuntu-0ubuntu2.7 ?

You can find binaries for it here:

https://launchpad.net/ubuntu/+source/samba/2:4.7.6+dfsg~ubuntu-0ubuntu2.7

(Click "Builds" on the right)

Thanks!

[Valentijn Sessink (valentijn)]

Done so. (FYI: upgrade came from 2.5, IIRC) 2:4.7.6+dfsg~ubuntu-0ubuntu2.7 installed now. I'll report what happens - if anything.

[Andreas Hasenack (ahasenack)]

Thanks, please keep us posted.

2.9: security fix (CVE-2019-3880)
2.8: made new API call available (bug #1778322)
2.7: memleak fix (bug #1814532)
2.6: smbc_opendir should not return EEXIST with invalid login credentials(bug #1801227)

[Valentijn Sessink (valentijn)]

So far, no more lockups. (And I'm almost sorry to say that, as I realize that it's not much we have here in this bug report - and it seems to only affect me...)

[Andreas Hasenack (ahasenack)]

That's ok, thanks for the update. Bugs are like that sometimes.

I'll mark this bug as incomplete in the meantime. Once there is new information, feel free to switch it back to new.

[Launchpad Janitor (janitor)]

[Expired for samba (Ubuntu) because there has been no activity for 60 days.]