Daniel, Thanks a lot for this playlist! Your videos helped a lot of people, including me. *Separate kudos for background music: ideal level, ideal style!
Thank you for sharing the video. I've built a Ceph cluster, every thing is working fine. Now I aim to create a file system, I followed the step in your video, but I got an error when I execute "systemctl start ceph-mds@node01". The error is: "System has not been booted with systemd as init system (PID 1). Can't operate. Failed to connect to bus: Host is down" Note that when I need to manage things in my ceph cluster I first do "sudo /usr/sbin/cephadm shell --fsid XX -c /etc/ceph/ceph.conf -k ". So I do it in the docker environment do you have any idea about the error message and how to fix it? My specs are: ceph: quincy host: ubuntu 20.04 docker version: 24.0.5 Thanks in advance
Hi Giuliano If you are using Cephadm you should not try to manually create MDSes. Then it's up to the automatic gods to give you services. Go into the services tab in the interface and choose to deploy 2-3 MDS services. I go through an installation and talk about setting up services in this video ruclips.net/video/4TLEg_OIU8M/видео.html I hope this helps. Thank you for watching my videos. Best regards Daniel
Hi, How can I verify if the data what we have copied on mon1 machine to /cephfs is also get replicated to the other mons machine? just wanted to verify if data is really replicated or not.
Hi Akshay Great question. The data isn't stored on the monitor per se. But they should be in sync, and each of them should know about the objects stored. Metadata should be replicated between the MDS, and the objects should be stored on different OSDs. To verify, we need to go down to the lowest level of the system. If we look at the rados object storage, we can read each object and see that they are stored on different filestores. Maybe this video will give you more insights: ruclips.net/video/sjE2QZ17uUo/видео.html If we store a really small file that only takes up one object. < 128Mb. Then we should see 3 objects stored in the metadata pool and 3 objects stored in the data pool. If we run `ceph pg dump` we get all placement groups in the system where we store data. Figuring out which PG has stored the data can be done with `ceph pg {pg-id} query` for each group and then see which OSD these are stored. I don't think this verification is essential for all data, as I expect the system to work. It has a long legacy of storing data, so much of the storage system has been verified and used for a long time by small and large systems. One of my favorite quotes from the CERN operator "If you have a small cluster with less than 10 Petabytes, then it should just work fine". They store massive amounts of mission-critical data in Ceph, so I feel safe storing my data there too. I hope this helps and that you have some information to proceed with your investigation. If you have any more specific questions, please don't hesitate to ask. Thank you for watching my videos. Best regards Daniel
Hi . How can I make subvolume with specific size. (100 GB) I want to make ceph fs filesystem subvolumes but after I mount filesystem i have one big size of cephfs filesystem in df -h command
Hi Piotr If you want to use the same resources / pool then it's easy to separate volumes for different uses. You create multiple directories in the root folder for example. Then you set a size quota on that folder and mount that folder on the machines. If security is important then you create separate users that have read / write access to the specific mount points. Thank you for watching my videos. I hope this helps. Best regards Daniel
Hi Angry Administrator Well you are right that the name limits me. But Persson is the 8th most common name in Sweden so that is where I'm from :) Best regards Daniel
@@DanielPersson True Swede? Do you like the taste and the smell of Surströmning? We once obtained a can. I twas the size of an anti-tank mine. After half a year in the fridge, noone of us daring touching it nor opening it, even though that was the ploan with the whole odeal, it started expanding. At grew by like 2 CM in the bottom and 2CM in the top. Now it looked like the land mine was ready to explode... We safely disposed of it, never experiencing neither smell nor taste. Abit dissapointing now that i think of it :/
Hi Angry Administrator. No, I've never tasted it. Smelled it, I have. Someone opened a can at Camp one year, indoors. We did not sleep there that night, and we had to leave early and were not allowed to return to that camp facility again. Then again, it was a leader that thought it would be funny; he just didn't know the poor thing :) Best regards Daniel
![Felix "Unfair" | [Stray Kids : SKZ-PLAYER]](http://i.ytimg.com/vi/Oswujxm2Ag0/mqdefault.jpg)
Daniel, Thanks a lot for this playlist! Your videos helped a lot of people, including me. *Separate kudos for background music: ideal level, ideal style!
Thank you Daniel, just Thank you :)
Thank you for sharing the video.
I've built a Ceph cluster, every thing is working fine. Now I aim to create a file system, I followed the step in your video, but I got an error when I execute "systemctl start ceph-mds@node01". The error is: "System has not been booted with systemd as init system (PID 1). Can't operate. Failed to connect to bus: Host is down"
Note that when I need to manage things in my ceph cluster I first do "sudo /usr/sbin/cephadm shell --fsid XX -c /etc/ceph/ceph.conf -k ". So I do it in the docker environment
do you have any idea about the error message and how to fix it?
My specs are:
ceph: quincy
host: ubuntu 20.04
docker version: 24.0.5
Thanks in advance
Hi Giuliano
If you are using Cephadm you should not try to manually create MDSes. Then it's up to the automatic gods to give you services.
Go into the services tab in the interface and choose to deploy 2-3 MDS services. I go through an installation and talk about setting up services in this video
ruclips.net/video/4TLEg_OIU8M/видео.html
I hope this helps. Thank you for watching my videos.
Best regards
Daniel
Hi, How can I verify if the data what we have copied on mon1 machine to /cephfs is also get replicated to the other mons machine? just wanted to verify if data is really replicated or not.
Hi Akshay
Great question. The data isn't stored on the monitor per se. But they should be in sync, and each of them should know about the objects stored. Metadata should be replicated between the MDS, and the objects should be stored on different OSDs.
To verify, we need to go down to the lowest level of the system. If we look at the rados object storage, we can read each object and see that they are stored on different filestores.
Maybe this video will give you more insights:
ruclips.net/video/sjE2QZ17uUo/видео.html
If we store a really small file that only takes up one object. < 128Mb. Then we should see 3 objects stored in the metadata pool and 3 objects stored in the data pool. If we run `ceph pg dump` we get all placement groups in the system where we store data. Figuring out which PG has stored the data can be done with `ceph pg {pg-id} query` for each group and then see which OSD these are stored.
I don't think this verification is essential for all data, as I expect the system to work. It has a long legacy of storing data, so much of the storage system has been verified and used for a long time by small and large systems.
One of my favorite quotes from the CERN operator "If you have a small cluster with less than 10 Petabytes, then it should just work fine".
They store massive amounts of mission-critical data in Ceph, so I feel safe storing my data there too.
I hope this helps and that you have some information to proceed with your investigation. If you have any more specific questions, please don't hesitate to ask.
Thank you for watching my videos.
Best regards
Daniel
Hi . How can I make subvolume with specific size. (100 GB) I want to make ceph fs filesystem subvolumes but after I mount filesystem i have one big size of cephfs filesystem in df -h command
Hi Piotr
If you want to use the same resources / pool then it's easy to separate volumes for different uses.
You create multiple directories in the root folder for example. Then you set a size quota on that folder and mount that folder on the machines.
If security is important then you create separate users that have read / write access to the specific mount points.
Thank you for watching my videos.
I hope this helps.
Best regards
Daniel
@@DanielPersson thank you very much . I did size quota . Best regards
Your accent and voice reminds me abit about Fyrstikken :D Norwegian I bet! Even though the name kinda limits it to a few select countries already.
Hi Angry Administrator
Well you are right that the name limits me. But Persson is the 8th most common name in Sweden so that is where I'm from :)
Best regards
Daniel
@@DanielPersson True Swede? Do you like the taste and the smell of Surströmning?
We once obtained a can. I twas the size of an anti-tank mine. After half a year in the fridge, noone of us daring touching it nor opening it, even though that was the ploan with the whole odeal, it started expanding. At grew by like 2 CM in the bottom and 2CM in the top. Now it looked like the land mine was ready to explode... We safely disposed of it, never experiencing neither smell nor taste. Abit dissapointing now that i think of it :/
Hi Angry Administrator.
No, I've never tasted it. Smelled it, I have. Someone opened a can at Camp one year, indoors. We did not sleep there that night, and we had to leave early and were not allowed to return to that camp facility again. Then again, it was a leader that thought it would be funny; he just didn't know the poor thing :)
Best regards
Daniel