-
-
Notifications
You must be signed in to change notification settings - Fork 3.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
A stop job is running for udev Kernel Device Manager - systemd 243.162-2 - initramfs delay #14128
Comments
I'm affected as well by this since the upgrade to systemd 243.162 |
cc @mwilck. |
Could you boot the system with |
I'm affected by the same issue too, after having updated systemd to version 243.162-2. Here, journalctl and dmesg logs with |
This is indeed related to my changeset #13961. However, please don't shoot the messenger. You've got 23 uevents that never finish: SEQNUM 1973-1979 and 2001-2017. This indicates that's something wrong with a few devices and/or drivers, perhaps you just never noticed so far. You have rebooted quickly, so that at the time the system is shutting down, some workers from the system startup haven't even finished yet:
So it seems that you have a problem with an ata device. You may want to hit
If you don't care about the root cause and just want to reboot more quickly, you can add |
Thanks for the detailed response @mwilck. It looks like you parsed @mcaliandro logs and not mine, which is fine. Also just to be clear this only happens on boot up for me, does not happen when shutting-down/rebooting. I do wonder though how you identify the "23 uevents that never finish". What is the best way to identify those queued actions that never complete? I'd like to go through my logs and see if I can figure out what device/driver is causing it for me specifically. Is it just finding all QUEUED events that don't also have a PROCESSED event? i.e Using @mcaliandro logs as an example, SEQNUM 2086 is queued in the first line, and then processed in the last line quoted below:
So I'd be looking for queued SEQNUMs that don't have a processed entry? Any insight is appreciated. You do mention using
So when I experience the 30s delay in my initrd when booting, just press the |
Yes, that's what I did. Here's the awk script that I used. However, it turned out in #14140 that some of these may be false positives due to lost messages. Please make sure that rate limiting is off in
Yes. You may need to boot with |
@CarbonChauvinist, you are not using the version with my changeset #13961. If you did, you wouldn't see the messages below:
In your case, the events that don't finish appear to be related to your SSD:
How is your root device set up? I can see LVM devices, are they mapped onto that SSD? |
@mwilck, obviously I defer to your expertise, but this seems strange to me:
Thanks for the awk script, I'll try that shortly. As for rate limiting/printk, I've done the following:
So the 860 is SATA SSD I have installed in my system, which contains a seperate Arch install. It does have LVM, but none that are being mounted to my new system installed to a M2/NVME 960. Not exactly sure what you want to see related to that, but here's the fstab from the system booting to the 960:
and a
On a positive note, I have confirmed that I can edit the But, I'll then get other delays for starting such as "Monitoring of LVM2 Mirrors", "Create Static Devices", and "Journal something or other", sorry they do start to quick to accurately write down. Also I've found that I need to edit Here's my latest boot dmesg and journalctl, because I'm sure I was clear as mud with all the above (though it looks like the dmesg didn't capture everything from the very beginning?). |
OK, let's get this straight: the arch package you are using does not contain the patch that I thought we were talking about: bfde942 ("udevd: wait for workers to finish when exiting"). It only contains 5db454b ("udevd: fix crash when workers time out after exit is signal caught"), which does not change udevd's runtime behavior, it just fixes a NULL pointer dereference. So, I dispute being responsible for this regression ;-) This does not mean that I'm unwilling to help. |
Oops, I should have read your text more carefully. So the culprit is 5db454b, wow. This patch does nothing but defer the dereferencing of the monitor from manager_exit() to manager_free(). Trying to understand how this could cause a hang... |
…aught" This reverts commit 5db454b. See systemd/systemd#14128
@CarbonChauvinist, please add 030f457 as backport or patch (keeping the revert of 5db454b in place), and try again. This would be equivalent to testing #14160. I believe the problem should not occur. |
@CarbonChauvinist You can check your /etc/fstab file. If you see something duplicated, just delete them. I have solved this problem this way. May you can try it. |
…aught" This reverts commit 5db454b. See systemd/systemd#14128 (cherry picked from commit 3cabdc2) [fbui: adjust context]
…aught" This reverts commit 5db454b. See systemd/systemd#14128 (cherry picked from commit 3cabdc2) [fbui: adjust context]
…aught" This reverts commit 5db454b. See systemd/systemd#14128 (cherry picked from commit 3cabdc2)
…aught" This reverts commit 5db454b. See systemd#14128
…aught" This reverts commit 5db454b. See systemd/systemd#14128 (cherry picked from commit 3cabdc2)
systemd version the issue has been seen with
Used distribution
Expected behaviour you didn't see
Unexpected behaviour you saw
Steps to reproduce the problem
The text was updated successfully, but these errors were encountered: