Problems with log file on SSD

Installing and Using OpenWrt
1

On my OpenWrt One I have moved the system.log to an SSD into the directory /work/logs. Every day at 23:59 a logrotate is started which removes a lot of unneccessary messages and copies the remaining ones to a daily /work/logs/system_DOW.log, where DOW is Mon, Tue etc.

After a reboot, the new messages are not written to /work/logs/system.log any more. A new logrotate after the reboot fixes the problem.

Maybe, the reason is the timing: Probably, logging is started before the SSD is mounted by /etc/rc.local.

Because an entry in the fstab did not work, I added the mount in the /etc/rc.local:

/etc/fstab

/dev/nvme0n1p1 /work btrfs subvol=@work,defauls 0 0

/etc/rc.local

# Mount the @work Subvolume permanently on the SSD
mount /dev/nvme0n1 /mnt/nvme
mount -o subvol=@work /dev/nvme0n1 /work
exit 0

Temporarily, I can do a restart of the logger directly after the mount in /etc/rc.local, but then I still lose messages before the mount in rc.local.

I would prefer a clean solution which mounts the SSD before the start of the logger if timing is reason for the loss of log messages.

Any ideas?

Netgear R6900v2 sometimes doesn't boot
2

I have now added a “service log restart” after the mount in /etc/rc.local but I have still lost the messages before the mount.

The question is now: How can I mount the (at)work btrfs subvolume as /work before logger is started?

3

Copy current log to /tmp
Stop logger
Mount disk
Copy log back from /tmp
Start logger

Yes, you'll miss some messages this way too, unfortunately.

4

Maybe it might be better if you could resolve the fstab error? Mounts are typically part of init, rc.local runs much later.

5

This has been resolved in the development branch but has not been backported to stabled. In the meantime, a copy of that version of /etc/init.d/log is here https://www.wildtechgarden.ca/doc/presentations/zabbix-on-openwrt/configure-database-storage-and-logging-for-zabbix/#configure-system-logging-to-srvlog

6

It seems to work now:

I have installed block-mount and created /etc/config/fstab:

config global
option anon_swap '0'
option anon_mount '0'
option auto_swap '1'
option auto_mount '1'
option delay_root '5'
option check_fs '0'

config mount 'work'
option target '/work'
option fstype 'btrfs'
option options 'subvol=@work'
option enabled '1'
option uuid '93cd1a03-22db-4a7c-bbd0-282b45d7c194'

The subvolume is mounted correctly and the result of a “dmesg | grep BTRFS” gives the following result:

[ 17.612778] BTRFS: device fsid 93cd1a03-22db-4a7c-bbd0-282b45d7c194 devid 1 transid 125885 /dev/nvme0n1 scanned by block (960)
[ 17.624905] BTRFS info (device nvme0n1): first mount of filesystem 93cd1a03-22db-4a7c-bbd0-282b45d7c194
[ 17.634351] BTRFS info (device nvme0n1): using crc32c (crc32c-generic) checksum algorithm
[ 17.642562] BTRFS info (device nvme0n1): using free space tree
[ 17.652374] BTRFS info (device nvme0n1): enabling ssd optimizations
[ 17.658649] BTRFS info (device nvme0n1): auto enabling async discard

7

This topic was automatically closed 10 days after the last reply. New replies are no longer allowed.