At large scale, we noticed that nova-compute was sending a lot of
messages to conductor.
Tuning the update_resources_interval parameter helped us drastically
reducing the load on rabbit and nova side.
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: Id4405691dcd4051ec64a448a82896c4f3e2a9594
The document mentionned that we could use
the rest of nodes as read only nodes, however the
configuration suggest to use only one.
Change-Id: I418cd0e0bd00b093356ecca3d949e5ad44ba260e
This proposal is to document a read-only haproxy configuration that can
be used by OpenStack services to lower the load on the master db
backend.
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: I6cf86c97ce1f2dd86d5da7c4e07384ae739ed10b
This patch propose a rework of the toctree so we have all configure
resources available in a subfolder instead of only one page.
This is also solving the issue introduced in 862100 (duplicate menu
entry).
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: I9b45309a981d0f0ddedc006b90499036b137360b
Removed FAQ section.
Moved memcached question to other resources.
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: I6a53f7ff50ae247e22440f50d33d28a32cfe8ca2
Move configuration guidelines under configure section instead of others.
The purpose behind this is to add some other configure/ pages (like one
for the database, neutron, nova, etc).
The fact that they were in Other resources was confusing to me.
Also separate resources from this SIG with external resources (links to
external websites).
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: Ifc1fba679bf5a816cd60b5aa3ae5a60ccf6c2492
We found a few interesting settings for databases in an IRC discussion.
This is the information we got out of it.
Change-Id: I7c16b32f6019bd5e2ee1343accd2d681d6f5adef
Depending on how erlang is started the options need to be passed with a
minus or a plus character. As this is not completely obvious we point
this out in our docs.
Change-Id: Iabedb7a132d743e40cb6253b1985076219128c84
The official rabbitmq dockerimage sets `+stbt db` as a launch option
of erlang. This option defines how the erlang scheduler threads (the
threads actually running the erlang code) are distributed over the
phyiscal cpu cores. Per default erlang spawns one scheduler thread per
logical core (so hyperthread or phyiscal core depending on if you have
hyperthreading enabled).
The setting `db` translates to the setting
`thread_no_node_processor_spread` which basically means that each
scheduler thread is pinned to one specific logical core in a
predetermined order. There are some more details to this specific
setting, but they are not relevant for this issue (if interested please
see https://www.erlang.org/doc/man/erl.html#+sbt).
When possible the erlang runtime tries to run processes on the lowest
possible scheduler id. As the scheduler id consistently mapping to a
logical core this means that if possible all rabbitmq processes on a
given host try to run code on the same logical core.
The erlang documentation already metions that this can cause
performance issues:
> If the Erlang runtime system is the only operating system process
that binds threads to logical processors, this improves the
performance of the runtime system. However, if other operating
system processes (for example another Erlang runtime system) also
bind threads to logical processors, there can be a performance
penalty instead. This performance penalty can sometimes be severe.
If so, you are advised not to bind the schedulers.
As this recommendation is quite hidden we add it to our config
recommendations as well.
The above is a summary of https://www.erlang.org/doc/man/erl.html#+sbt.
Change-Id: Iee95287911097274c96c630fa912e7f994339b6a
The How to Contribute page was incomplete and a bit off since it
was designed for code repositories.
Change-Id: I22481829230b8e202b9b1d43dd7b5a9a2ae21fd1
Clean up main index and add all missing information from the wiki
page, so that we can start discontinuing the wiki content.
Change-Id: I250f0441994811d3ff949c08f81ecface53783a9
This params should be tuned correctly when running large-scale.
The default value is 30 minutes.
My recommandation is to set this to 60secs.
Signed-off-by: Arnaud Morin <arnaud.morin@ovhcloud.com>
Change-Id: I3fa643ef94a60474273902e95d4e03b9f31164d4
As part of refactoring the documentation adding the files that will be filled in the later patches.
This is needed for easier linking between the files.
Signed-off-by: Ramona Rautenberg <rautenberg@osism.tech>
Change-Id: I365a66d1bb6191c114352fd8093a4eca453fb1e8
Created as a combination from a trimmed down cookiecutter run and some
copying from the arch-design repo.
Change-Id: I8e4038337394c861e9282fe7e238977930d9db4b
