etcd on unstable or unsupported architectures will print a warning message and immediately exit if the environment variable ETCD_UNSUPPORTED_ARCH is not set to the target architecture.
Change-Id: Ifdd3ed4ae322484477d2db3e2d0eed09e086518f
Signed-off-by: Jiamei.Xie <Jiamei.Xie@arm.com>
* Add initial support for capabilities
This eliminates the need to set MYSQL_FLAVOR
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Removed MYSQL_FLAVOR variable precedence
It will cause backwards compatibility problems,
since many users set "MySQL56" for any MySQL system.
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Address PR feedback (WIP)
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Fixed mysql_safe to not use capabilitities
Reverted small changes to configs
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* fix version detection
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Fix style, re-add EXTRA_MY_CNF from review
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Address review feedback
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* address remaining feedback
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Remove rogue added space
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Address PR feedback
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Address PR feedback
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Fix broken test
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Fixed tests
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Addressed PR feedback
Added tests for version detection
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* update copyright
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Remove unset MYSQL_FLAVOR
I thought about this: it's a bad idea.
It means that if there was a test that created 2 mysqld's, the usage
would not be repeatable/idempotent.
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Make sure error is caught.
Signed-off-by: Morgan Tocker <tocker@gmail.com>
* Address PR Feedback
Signed-off-by: Morgan Tocker <tocker@gmail.com>
The error from bootstrap.sh was:
"./dev.env: line 23: source: build.env: file not found".
I think it broke in commit 157ebf5830 .
Not sure if this is the right fix, but it worked for me
(bootstrap.sh does `source ./build.env`, so it's probably ok).
Signed-off-by: Scott Lanning <scott.lanning@booking.com>
dev.env is sourced from the shell and for some build environments
like rpm the shell used for build "snippets" MUST be a posix
shell, not bash.
So adjust one occurence where trying to build a vitess rpm from
rpmbuild failed due to bashisms not recognised by the posix shell.
Also remove the shebang line as this script is not run in
isolation but sourced from an existing shell.
Signed-off-by: Simon J Mudd <sjmudd@pobox.com>
This is tracked here: https://github.com/youtube/vitess/issues/3219
Note that this also triggers a Debian version upgrade to their latest
stable version "stretch" which was released this summer.
This required several additional changes and cleanups:
- Remove outdated GO15VENDOREXPERIMENT.
- Docker: Upgrade Java to 8 (from 7).
- Docker: Upgrade PHP to 7 (from 5).
- Docker: Use "pip" from distribution and do not install it ourselves.
- Docker: PHP: Use "pecl install grpc" and no longer compile it ourselves. Relevant parts in travis/install_grpc.sh have been removed.
- Docker: Do no longer install etcd and consule in the common image because we call bootstrap.sh when building the flavored image and that installs these two as well.
- travis/install_grpc.sh changes:
- Do no longer compile gRPC because for PHP and Python we can use other ways to install the shared library.
- Do no longer compile protobuf. Instead, download pre-compiled "protoc" binary in bootstrap.sh.
- Updated Python and PHP versions to 1.6 as we did earlier for Go.
Please refer to #2694 and #2670 for motivation and reasoning for
this change.
I've tried to follow best practice in inserting the copyright
headers. In other open source projects, not all files carry
the notice. For example documentation doesn't. I've followed
similar ground rules.
I did not change the php because there is a separate LICENSE
file there by Pixel Federation. We'll first need to notify
them our intent before changing anything there.
As for the presubmit check, it's going to be non-trivial
because of the number of exceptions, like file types,
directories and generated code. So, it will have to be
a separate project.
This gets rid of the warning about disabling native vendor mode
due to legacy Godeps/_workspace directory being present.
It also means we require Go 1.5+ since that's when native vendor
mode was first introduced.
It should match the convention established with "MariaDB".
I looked into making them case-insensitive, but it would be more tricky
than it's worth because flavor IDs are used as part of the protocol
between servers. So an old server that is case-sensitive would choke on
a message from a new one.
Morgan Tocker
Jiamei.Xie
Arindam Nayak
yuxiaobo
Scott Lanning
Jitendra Vaidya
Sugu Sougoumarane
Simon J Mudd
Michael Berlin
Ashe Connor
Anthony Yeh
Alain Jobart
Rafael Chacon
Simon Mudd
Alex Charis
acharis
thompsonja
Joshua Thompson
Ammar Aijazi
Shengzhe Yao