Hey folks, ceph is a large system, and like any large system testing it
can be quite complex.
To help new and existing developers I'd like to organize some training
sessions on different aspects of testing, and record these for
posterity.
We can start out with basics, like how to run tests, and then cover more
complicated topics like how to develop new teuthology tasks, or cephadm
integration. This can also be a useful source of content for updating
the developer docs, like Deepika Upadhyay is doing here:
https://github.com/ceph/ceph/pull/37949
Here's an etherpad to capture topics to cover - please take a look and
add more:
https://pad.ceph.com/p/teuthology-topics
Josh
There is a general documentation meeting called the "DocuBetter Meeting",
and it is held every two weeks. The next DocuBetter Meeting will be on
January 27, 2021 at 0200 UTC, and will run for thirty minutes. Everyone
with a documentation-related request or complaint is invited. The meeting
will be held here: https://bluejeans.com/908675367
THIS MEETING IS 9 PM WEDNESDAY IN NEW YORK.
THIS MEETING IS 7 PM WEDNESDAY IN CALIFORNIA
Send documentation-related requests and complaints to me by replying to
this email and CCing me at zac.dover(a)gmail.com.
The next DocuBetter meeting is scheduled for:
27 Jan 2021 0200 UTC
Etherpad: https://pad.ceph.com/p/Ceph_Documentation
Meeting: https://bluejeans.com/908675367
Thanks, everyone.
Zac Dover
Hi everyone,
With Pacific nearly out the door, it's past time to finalize the name for
the Q release. Traditionally we have chosen cephalopod species. For Q,
the options are more limited, so we've had to stretch what qualifies as a
cephalopod species name a bit. The contenders:
Quincy: for Squidward Quincy Tentacles of Spongebob fame
Quebecoceras: an extinct genus of nautiloid
Quarren: a species of aquatic humanoids (nicknamed "squid heads") from the
Star Wars universe
Qarnit: Maltese for "octopus"
There were a few non-cephalopod suggestions, notably Quarantine.
Please vote here by adding a "+1" to your preferred name:
https://pad.ceph.com/p/q
Thanks!
sage
Hello,
I am having an error while trying to run rados benchmark after running
vstart script
I run :
../src/vstart.sh -d -n -l
and then when I try to run:
bin/rados -p rbd bench 30 write
it gives me error saying:
error opening pool rbd: (2) No such file or directory
Can someone please help me in this. I will be grateful
Thanks
Hi there,
I am Shivam a final year student from PICT, Pune.
I am looking forward to contribute to ceph and aiming for gsoc(ceph must be
applying as well as org).
To introduce myself in short I have hands-on experience of working on C++
and Rust by working as an intern at Igalia(Chromium browser) and Linux.
I have a good command of algorithms and problem solving using data
structures.
I tried joining irc channel but it seemed like i went in some other
channel, please help me out.
At the moment i have cloned ceph repo and I am building it.
I've removed the containers from the cephci organization on quay.io;
cephci builds are now stored on quay.ceph.io and have been for some
time, and the presence on quay.io was just confusing and pointless.
Every build I removed was over a year old.
Hi Folks,
The weekly performance meeting is starting in 15 minutes! Today I will
talk a little about some early crimson testing with fio and we may
discuss on-going performance work with the classical OSD as well.
Please feel free to add your own topic as well.
Hope to see you there!
Etherpad:
https://pad.ceph.com/p/performance_weekly
Bluejeans:
https://bluejeans.com/908675367
Thanks,
Mark
Dear Daniel and those familiar with Project Zipper,
I am interested in File backend mentioned in the following email, which is part of Project Zipper.
https://marc.info/?l=ceph-devel&m=155542425904339&w=2
Daniel, do you have any idea in mind about how it will be implemented?
For example, what kind of file interface (e.g. POSIX, NFS) will RGW use to access the backend?
I am interested in File backend because it might be useful if RGW could be backed by File backend such as NFS server instead of RADOS.
I guess RADOS does not work well on some storage devices such as tape, but such devices can alternatively be accessed via file interface.
In such a case, it might be useful if RGW could access the File backend and treat it as a different storage class from STANDARD.
Sincerely,
Ken Iizawa
Hello,
I am studying the ceph paxos implementation.
In handle_last, the leader counts the replies to collect phase by incrementing the variable num_last and without checking if the reply is duplicated (however, in handle_accept, the leader keeps a set of peons that replied, in variable accepted).
Why does the leader not keep a set of peons that replied in the collect phase, similar to what it does later in function handle_accept?
Thanks,
Afonso
Hi all,
I am trying to build master on 32 bit ARM machine. cc1plus gives up with
"virtual memory exhausted: Cannot allocate memory" when trying to build
unittest_librbd with Release build type (not even Debug - so no debug
symbols). I see cc1plus process grow to 3GB (maximum userspace available
under Linux on 32 bit machines) and then just bailing out (presumably
when it is unable to alloc more memory). unittest_librbd consists out of
number test suites and running them all together is a tall order for 32
bit machines. In fact TestMigration is so big so it cannot successfully
pass without bad_alloc even if it is built on 64 bit machine for 32 bit
system. So I suggest to break unittest_librbd to number of separate test
suites to avoid issues with building and test execution. I am willing to
contribute by refactoring it but want to get devs view of the problem
before doing so.
Regards,
Vladimir
