| View previous topic :: View next topic |
| Author |
Message |
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
 Posted: Wed Jan 10, 2024 3:15 pm Post subject: run-crons reporting missing file |
 |
|
A couple days ago several machines suddenly reported the following (through system email)
| Code: | Subject: Cron <root@ng2> test -x /usr/sbin/run-crons && /usr/sbin/run-crons
...
Opening file "/tmp/conftest91" failed, ignoring: No such device or address |
I have tried getting to the root-cause of this, but to no avail unfortunately.
| Quote: | acct-group/cron-0-r1
acct-group/crontab-0-r1
acct-user/cron-0-r2
sys-process/cronbase-0.3.7-r10
sys-process/cronie-1.7.0-r2
virtual/cron-0-r3
|
should i be worried about this? any idea on a way to determine the reason for these messages ?
cheers,
guid0 |
|
| Back to top |
|
 |
Hu
Administrator
Joined: 06 Mar 2007
Posts: 21741
|
| Posted: Wed Jan 10, 2024 4:23 pm Post subject: |
|
|
| What is /tmp/conftest91? I have historically seen some package builds leave garbage in /tmp, always with a name following the shown format. Perhaps this is a socket, block device, or other pseudo-file left over from a configure test that failed to clean up after itself. |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 4:45 pm Post subject: |
|
|
I really have no idea what /tmp/conftest91 is  and the file is not present. What i do know is that its being reported on 7 machines which were installed over a few years. Some machines have been running for a while now (few years) and only gotten kernelupgrades / microcode when security demanded it.
Checked sockets, pids, ports and major/minor device numbers, none have 91..
really no clue what this could be.
| Code: | # rc-update
binfmt | boot
bootmisc | boot
cgroups | sysinit
cronie | default
devfs | sysinit
dmesg | sysinit
fsck | boot
hostname | boot default
hwclock | boot
ip6tables | default
iptables | default
keymaps | boot
killprocs | shutdown
kmod-static-nodes | sysinit
local | default nonetwork
localmount | boot
loopback | boot
mcelog | default
mdadm | boot
modules | boot
mount-ro | shutdown
mtab | boot
net.eth0 | default
net.eth1 | default
net.eth2 | default
net.eth3 | default
netmount | default
nginx | default
nrpe | default
ntpd | default
postfix | default
procfs | boot
root | boot
save-keymaps | boot
save-termencoding | boot
savecache | shutdown
seedrng | boot
sshd | default
swap | boot
sysctl | boot
sysfs | sysinit
sysklogd | default
systemd-tmpfiles-setup | boot
systemd-tmpfiles-setup-dev | sysinit
termencoding | boot
udev | sysinit
udev-trigger | sysinit
|
|
|
| Back to top |
|
|
Christian99
Veteran
Joined: 28 May 2009
Posts: 1673
|
| Posted: Wed Jan 10, 2024 5:38 pm Post subject: |
|
|
| maybe you can check the /etc/cron.*/ dirs which script actually tries to open the file and see why, and why it fails. |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 6:36 pm Post subject: |
|
|
I checked below scripts (not that many) using strace and none show any interaction with tempfiles. The /etc/crontab is near empty and only a wrapper for the periodic runs. There are no user crontabs.
| Quote: | /etc/cron.monthly/smartmontools-update-drivedb
/etc/cron.weekly/mdadm
/etc/cron.hourly/0anacron
/etc/cron.daily/logrotate
/etc/cron.daily/man-db
/etc/cron.daily/systemd-tmpfiles-clean |
/etc/conf.d/cronie
| Code: | # Settings for the CRON daemon.
# CRONDARGS= : any extra command-line startup arguments for crond
CRONDARGS= |
|
|
| Back to top |
|
|
grknight
Retired Dev
Joined: 20 Feb 2015
Posts: 1693
|
| Posted: Wed Jan 10, 2024 6:59 pm Post subject: |
|
|
Is /tmp mounted in a special way? Like set up as a tmpfs or zram?
This sounds like some sort of backing error where the main run-crons tried to claim a temporary file but something blocked it. |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 7:54 pm Post subject: |
|
|
nothing really special afaik: | Code: | | /dev/sda5 on /tmp type ext4 (rw,nodev,noexec,noatime) |
|
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 7:58 pm Post subject: |
|
|
on one of the other machines its mounted slightly different:
| Code: | | /dev/md2 on /tmp type ext4 (rw,nosuid,nodev,noexec,noatime) |
still nothing really special.. i do appreciate all the new angles to look at this. |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 8:08 pm Post subject: |
|
|
hmm, i might be on to something as there is below message in a stale maildir at the same time (not day) as the conftest91 messages.
| Code: | error: state file /var/lib/misc/logrotate.status is already locked
logrotate does not support parallel execution on the same set of logfiles. |
|
|
| Back to top |
|
|
grknight
Retired Dev
Joined: 20 Feb 2015
Posts: 1693
|
| Posted: Wed Jan 10, 2024 8:12 pm Post subject: |
|
|
| guid0 wrote: | nothing really special afaik: | Code: | | /dev/sda5 on /tmp type ext4 (rw,nodev,noexec,noatime) |
|
I am guessing from the failed name that a script is creating a "configuration test" and trying to execute it on /tmp. Since it is mounted as noexec, it fails.
Did this change recently as to coincide with noexec? |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Wed Jan 10, 2024 8:18 pm Post subject: |
|
|
| grknight wrote: | | Did this change recently as to coincide with noexec? |
noexec on /tmp has been part of my Gentoo-setups for more than 20+ years  |
|
| Back to top |
|
|
guid0
Guru
Joined: 06 Jul 2003
Posts: 371
Location: The Netherlands / Nederland
|
| Posted: Tue Apr 30, 2024 12:49 pm Post subject: |
|
|
I just happened to see the file reappear on a machine.
| Code: | # ls -laF /tmp/conftest91
srwxr-xr-x 1 portage portage 0 Apr 30 12:16 /tmp/conftest91= |
Here are the commands that were run.
| Code: | emerge --sync && emerge -upvDN world
uname -a
eselect news list
eselect profile list
emerge -upvDN world
emerge -uvDN world (quickly followed by a CTRL-C)
emerge -upvDN glibc
emerge -uvDN glibc
emerge -upvDN world
emerge -upvDN world
emerge -uvDN world
emerge -upvDN world
exit
|
Could this be an internal file to a portage activity and leaving it behind when hitting CTRL-C mid-process? |
|
| Back to top |
|
|
Hu
Administrator
Joined: 06 Mar 2007
Posts: 21741
|
| Posted: Tue Apr 30, 2024 1:55 pm Post subject: |
|
|
| Assuming that this file is written by an autoconf test, then yes, interrupting it partway through with Ctrl-C could cause the file not to be deleted properly. Many packages that Portage can build use autoconf / automake in their build process, so an emerge @world could have been running an autoconf-generated script at the time you interrupted it. |
|
| Back to top |
|
|
|
Portage & Programming
