Skip to main content
Topic solved
This topic has been marked as solved and requires no further attention.
Topic: Fatal error - read only (Read 966 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Fatal error - read only

Hello, and thank you for implementing S-6 in our lives.

 I have many fatal error - read only informations at boot.
 Where is that logs stored so I can paste it ?
 Thank you.

sudo fsck.ext4 -C0 /dev/sda8

 System is clean.
 


Re: Fatal error - read only

Reply #2
I have many fatal error - read only informations at boot.
 Where is that logs stored so I can paste it ?
 

It should be in /run/uncaught-logs/current. What's most likely happening here is that your root is initially mounted as read-only. I honestly don't know why this occurs to some people, but it can happen. Most s6-rc logruns try to write logs to /var/log when they start. The boot sequence is all parallel, so you will have s6-rc longruns immediately try to write to /var/log which is initially read-only and thus fail. s6-supervise should just restart them until they work again (after the root is remounted read/write) and you should good to go.

I don't want to artificially make these daemons wait for people that do have read/write initially (should be the vast majority), but it would be nice to eliminate these types of messages to avoid confusion. I guess what I could do is devise a nice generic way to initially log in /run if necessary (that will always be writable) and then switch it to /var/log after remount-root is done. The udevd longrun currently works like this.

Re: Fatal error - read only

Reply #3
@4000000060a590a1148ca4d3 install: cannot create regular file '/var/log/wtmp': Read-only file system
@4000000060a590a114b89503 install: cannot create regular file '/var/log/btmp': Read-only file system
@4000000060a590a117f99e36 s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a1188f9c2e s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a11c210d24 s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a20970f189 s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a2097d7b22 s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a20997b3cc s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a30923397e s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a309253ed6 s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a30954e7b7 s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a40938aaee s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a502ca61f7 s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a502ca7381 s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a5095869d4 s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a5096019c6 s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a509c0bea1 s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a60c96591e s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a60d3c8235 s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a60d9b0b68 s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a70ba4f145 s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a70bae536c s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a70be2d65d s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a80adce2eb s6-log: fatal: unable to open /var/log/NetworkManager/lock: Read-only file system
@4000000060a590a80b1bac46 s6-log: fatal: unable to open /var/log/dbus/lock: Read-only file system
@4000000060a590a80b97455d s6-log: fatal: unable to open /var/log/dmesg/lock: Read-only file system
@4000000060a590a8109b7d57 fuse: failed to access mountpoint /mnt/CB87A2E532F268E1: No such file or directory



So it seems like noveau error :)
 Just kidding, I am remembering noveau error that boots up every time I used other distributions.
 It means that's everything fine ?

Re: Fatal error - read only

Reply #4
The s6-log failures are expected in your case. I assume you log in and everything as normal. So yeah, those are fine. They're just noise. They get started properly later. That being said there's 3 errors there that stick out.

Quote
@4000000060a590a1148ca4d3 install: cannot create regular file '/var/log/wtmp': Read-only file system
@4000000060a590a114b89503 install: cannot create regular file '/var/log/btmp': Read-only file system
@4000000060a590a8109b7d57 fuse: failed to access mountpoint /mnt/CB87A2E532F268E1: No such file or directory

I'm not sure what's causing these messages.

 

Re: Fatal error - read only

Reply #5
I guess what I could do is devise a nice generic way to initially log in /run if necessary (that will always be writable) and then switch it to /var/log after remount-root is done. The udevd longrun currently works like this.

Just an update, I implemented this feature for all of our *-log daemon scripts. If /var/log is read only for whatever reason (the kernel doesn't like your filesystem or something), the logger daemons will instead write in /run/log (which is always writable). In the remount-root oneshot, a helper script is executed after the root partition is mounted read/write. This moves all log saved in /run/log to the appropriate /var/log directory and restarts the logger daemons so they point to /var/log instead. This should eliminate those s6-log message on bootup and also start daemons a wee bit faster for people that have their root initially mounted as read-only instead of read/write since it doesn't have to wait on udev to settle.

If your root is initially mounted as read/write already (should be most of you), nothing changes.