Posts by steakhutzeee

Hi,

i noticed this entry in my syslogs:

kernel: [1816962.694777] TCP: enx001e06328f28: Driver has suspect GRO implementation, TCP performance may be compromised.

Any idea what it is and if i have to do something?

Thanks!

Quote from ryecoaaron

I assume you mean a different server? Yes.

Yep. I'm thinking on a smart solution to automatically move the backup to the cloud. Maybe a cron job with rclone.

Quote from ryecoaaron

ddfull gives you an image you can just write to the media. dd requires you recreating grub (on some systems), mbr, etc, and then write the partition image. So, ddfull is just easier. I kept the dd option because some people have extra data partitions that might be very large that they don't want in the OS backup.

Thanks a lot!

So my first backup took 40 minutes. All 32 GB were read and the dd full backup image is less than 2 GB. The microsd is 32 GB. Cloning it after shutdown only took 8 minutes. It's because it's live?

Two other things:

-It's ok to manually delete the backup files i assume.

-i can just copy somewhere else the .ddfull.gz file ignoring the other files generated by the backup process? If yes, why they are needed also after the backup is complete?

What is the difference between using dd to backup the os partition and dd full to backup the drive?

Based on the fact there is only the os on my microsd card.

Quote from ryecoaaron

That is what I would do.

Maybe this is dumb but i'll try, say i restore and it seems to be working. There is some way to check the system integrity. Just to be sure that issues that could happen after some time are not related to the restore.

Quote from ryecoaaron

It is better to shutdown. But what I am trying to say is you keep backing up your system by shutting it down say once a month but use the online backup daily, I think you can get the best of both worlds.

So in case of disaster the first restore should be from the ddfull and in case it does not work as intended i can use the backup performed shutting down?

Quote from ryecoaaron

For the average home user, you would never notice. This is also why I prefer copying at the file level. Your backup may have older files but not partially changed blocks.

So you say it's better to shut down and clone right?

Quote from ryecoaaron

If you choose ddfull, the plugin runs a dd command that will read the blocks of the os drive sequentially and write them to a file. If one of those blocks changes after it has been read and written to the image file, it could cause an issue.

So this could happen:

dd reads a block, the block change in the meantime, the read value is written - but the block changed in the middle so it's wrong, Correct

Well powering down and clone the microsd sounds a lot safer. Assumed that the user should always test the backups, the ddfull seems a risk.

Quote from ryecoaaron

It runs a command against the system. If the data changes while the image is being written, then it will be wrong when restored. Most of the time (databases like mysql excluded), it won't cause a problem.

Interesting, could you elaborate more?

Or better, maybe i can do some kind of check before starting a backup of this kind?

Thanks for your reply.

On my microsd i run the system and nothing else iirc. Also docker path is on my HD and not on the microsd.

Hi,

was reading this thread [Backup Plugin] create a single img backup image for use with win32 disk imager / etcher, etc.

So if i understand this correctly, i can install the openmediavault-backup plugin and:

-backup full microsd card.

-write the resulting image to a microsd card to restore.

Got it right?

I'm actually powering down my Odroid HC2, removing the microsd and cloning it with usbimager.

Would like to ask how the plugin works, or better saying, how backing up a running system works in this case? Should i stop some services before starting the backup process?

Thanks in advance!

So i tried with this image: https://hub.docker.com/r/gists/dnsmasq

Port 53 was not available so i used 54:

version: "2.1"
services:
  dnsmasq:
    image: gists/dnsmasq
    container_name: dnsmasq
    #cap_add:
      #- NET_ADMIN
    ports:
      - 54:53/tcp
      - 54:53/udp
    volumes:
      - /srv/dev-disk-by-label-HC2/AppData/dnsmasq/etc/dnsmasq.d:/etc/dnsmasq.d
    restart: unless-stopped
  
networks:
    default:
      external:
        name: my-net

Inside the bind mount /etc/dnsmasq.d i created the file dnsmasq.conf with this content:

address=/dk.lan/192.168.1.197
#resolv-file=/etc/resolv.dnsmasq
listen-address=192.168.1.197
listen-address=127.0.0.1
bind-interfaces

No luck, the dnsmasq container logs says:

dnsmasq: failed to create listening socket for 192.168.1.197: Address not available

So i tried to look after some arm compatible and up to date dnsmasq docker image but found nothing suitable.

Is actually safe to install dnsmasq directly on the OMV host?

Hi,

I setup a Traefik container to reverse proxy my other containers locally and access them with "container.dk.lan", instead of "dk.lan:container_port".

dk.lan is my omv hostname.

Actually this is not working because i think I should add dns records in order to access "container.dk.lan".

Any advice I could I do this in OMV?

Thanks in advance!

Quote from ryecoaaron

The file really should be re-created because some of them are just wrong. On your system, waitAllMounts.conf should just be:

[Unit]

After=local-fs.target srv-dev\x2ddisk\x2dby\x2dlabel\x2dHC2.mount

then systemctl daemon-reload

Display More

Well at this point i could edit the file as you mentioned and reload the daemon. Could work in theory?

Quote from ryecoaaron

I never said using the delay won't work. I just don't like it because it could be waiting time it doesn't need to or not wait long enough. Waiting for all of the mounts (which is what the code is supposed to do) should work but I give up. Listen to someone who uses the setup that you are trying to get working (hint: not me).

Never said that, mentioned in fact that you said it's not optimal.

The point i'm missing is, the mounts number in the waitAllMounts.conf file depends on the number of the running containers? Or i can stick with the file i created?

Quote from KM0201

Then you did something wrong.. Multiple people have done exactly what was in that thread w/ zero issues.

Probably is like he said before:

"The override file is created when docker is installed by omv-extras. Running the commands after docker was installed with containers running added too many mounts."

Quote from KM0201

If you really believe your drive is not mounting until after docker starts (I've personally never had that issue w/ an HC2 internal drive), after the next "blackout", just execute the command

Code

 systemctl restart docker

then after a minute or so, if you're right your dockers should be up like normal. If that's the case, then you have correctly diagnosed the problem.

These are exactly the same steps i followed, as i said in the first post.

Quote from KM0201

To fix it...

If it were me, I'd delete that conf file, and give this thread a read...

Jellyfin portainer stack completely reset after 10-hour server shutdown: why?

The solution provided in the link you posted is the same as the first one proposed in the link i posted in my first post. This solution is not optimal as ryecoaaron says. Instead he suggests to use the waitAllMounts.conf file. The same file i issued this thread about.

Quote from ryecoaaron

The override file is created when docker is installed by omv-extras. Running the commands after docker was installed with containers running added too many mounts. But either way, it doesn't really matter because those determine when docker starts. Are you using restart-always for the containers?

Hi, thank you for your reply.

I use restart: unless-stopped.

By your response it's not clear to me if the .conf file generated is ok that way or if i should delete it.

No one?