I have been working on a new open-source project called Argonaut. I
would
love to get feedback from the Ceph community.
Looking at the README and "commit history" this kind of screams vibe
coded which turns me off from even trying to use it tbh. If you used
LLMs you should be transparent about it.
Argonaut is a terminal UI for administering Ceph clusters. The
inspiration
comes from tools like k9s and lazydocker: a keyboard-driven interface
for
common operational tasks while remaining transparent about what is
happening under the hood.
Sounds great in concept! Especially being transparent about what
happens.
Some questions I would like to ask you:
- Which operational workflows consume the most time today?
- Which information do you find difficult to discover quickly?
- Which features would make a TUI genuinely useful in your daily
workflow?
Most admins probably have (shell) scripts for most actions so answering
is kind of hit or miss. Some actions we have scripts for that take some
time.
Many of the things listed below are possible/easier over the dashboard.
- Listing connected clients
RBD, RGW, CephFS, ... every client connected is listed through a
different interface. If I quickly want to check who is using what
service right now I have to use my scripts.
- Actual ETA for rebalance/recovery
What it says on the tin. We got a script that checks how many objects
need to be moved, are moved per second and depending on the remaining
count map it to a curve to guesstimate an ETA
- Diagnosing OSD problems
This is probably a me problem, but:
Sometimes OSDs hang intransparently (they are marked as up, react to
status and even are able to be benched, but don't work and block the
cluster) would be nice to test them using a tui instead of my script
- Bucket usage/User usage by bucket
we have automated usage statistics by user/bucket - it's important that
the TUI supports different zone/realms
Dumping everything to a json file or something like that would be a plus
- managing data movement
When doing maintenance, adding hosts, replacing OSDs and so on we often
manage the data movement using scripts/by hand to minimize data
movement.
Doing this using a tui would be interesting.
- better PG overview
This is not something that normally takes up much time, but mostly
because we do it seldom. We got a multi site cluster which (without
extra patches) limits our usage of the autoscaler (it doesn't work at
all without our patches)
being able to browse pools by pg_num/size with some usage info
Like I said got a heap of tools of my own and am using scripts like
https://github.com/TheJJ/ceph-balancer as well.
For Ceph newcomers something that would do something like the
https://analyzer.clyso.com/ could be neat. Ceph warns you if you do
really stupid things, but some tool that tells you if you did slightly
stupid things or could optimize your setup could be neat.
Anyway most of that should be relatively specific to my cluster and I
probably forgot things I do by muscle memory, but I hope I at least gave
you some inspiration.
If anything what I said is wrong or could be done easily with the tools
available I would be glad if someone pointed me in the right direction
as well.
Kind regards
Thomas
On 2026-07-11 17:32, Sinan Polat via ceph-users wrote:
Hi everyone,
I have been working on a new open-source project called Argonaut. I
would
love to get feedback from the Ceph community.
Argonaut is a terminal UI for administering Ceph clusters. The
inspiration
comes from tools like k9s and lazydocker: a keyboard-driven interface
for
common operational tasks while remaining transparent about what is
happening under the hood.
It is still an early release and I am particularly interested in
feedback
from operators who manage production clusters.
Some questions I would like to ask you:
- Which operational workflows consume the most time today?
- Which information do you find difficult to discover quickly?
- Which features would make a TUI genuinely useful in your daily
workflow?
The project is available here: https://github.com/cinpol/argonaut
Any feedback, feature requests, bug reports, or criticism would be
greatly
appreciated.
Kind regards
Sinan Polat
_______________________________________________
ceph-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
--
Thomas Schneider
IT.SERVICES
Wissenschaftliche Informationsversorgung
Ruhr-Universität Bochum | 44780 Bochum
Telefon: +49 234 32 23939
http://www.it-services.rub.de/
_______________________________________________
ceph-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
