upgrade to manticore and RHEL9 #11
Conversation
akostadinov
force-pushed
the
manticore
branch
3 times, most recently
from
0926560 to
f7030f9
Compare
akostadinov
force-pushed
the
manticore
branch
from
f7030f9 to
dac7a35
Compare
akostadinov
force-pushed
the
manticore
branch
8 times, most recently
from
2e66084 to
47f6c18
Compare
akostadinov
force-pushed
the
manticore
branch
6 times, most recently
from
b303363 to
d7764d8
Compare
Containerfile
Outdated
| CC=clang-16 \ | ||
| CXX=clang++-16 \ | ||
| BUILD_PATH=/tmp/manticore_uselessly_long_path_to_prevent_rpm_build_issues \ | ||
| BUILD_FLAGS="-DUSE_SYSLOG=0 -DWITH_GALERA=0 -DWITH_RE2=0 -DWITH_STEMMER=0 -DWITH_ICU=1 -DWITH_SSL=1 -DWITH_ZLIB=1 -DWITH_ODBC=0 -DWITH_EXPAT=0 -DWITH_ICONV=1 -DWITH_POSTGRESQL=0 -DWITH_MYSQL=0 -DBUILD_TESTING=0" |
There was a problem hiding this comment.
Just wondering, are any of the BUILD_FLAGS set to some default values, or do we need to explicitly define all of them?
There was a problem hiding this comment.
not sure about -DWITH_ICONV=1 and -DUSE_SYSLOG=0 but I find it useful to see the settings on a glance
There was a problem hiding this comment.
It's not documented what are the defautls, so I prefer to keep them. But now I notice that icu can be used from RHEL too with -DWITH_ICU_FORCE_STATIC=1
Containerfile
Outdated
| chmod 770 /var/lib/searchd /var/run/sphinx | ||
| COPY --from=builder /tmp/manticore_uselessly_long_path_to_prevent_rpm_build_issues/build/*.rpm /tmp/rpms/ | ||
| COPY --from=$PORTA_IMAGE /opt/system/config/standalone.sphinx.conf "/etc/manticoresearch/manticore.conf" | ||
| ENV MANTICORE_RPMS="manticore-converter* manticore-icudata* manticore-common* manticore-server-core* manticore-server*" |
There was a problem hiding this comment.
Why not just manticore-* ? or would it contain something that we don't need?
There was a problem hiding this comment.
Yes, some RPMs have deps that are not satisfied. And we don't really need them.
Containerfile
Outdated
| rpm -iv --excludedocs $MANTICORE_RPMS && \ | ||
| cd - && rm -rf /tmp/rpms && \ | ||
| microdnf clean all && \ | ||
| sed -i -e 's#/var/run/sphinx/#/var/run/manticore/#' -e 's#/var/lib/searchd#/var/lib/manticore#' /etc/manticoresearch/manticore.conf && \ |
There was a problem hiding this comment.
I guess this is temporary, and when it's all finalized and deployed, we can just update the paths in the productized build?
Maybe we can leave a #TODO here for this task?
There was a problem hiding this comment.
Would be nice to do at some point. I don't consider it a big issue though. If you find it important I can add a TODO.
There was a problem hiding this comment.
Yeah, I think it would be nice to add a TODO, so we don't forget to do this eventually.
Not very important indeed, but it would be cleaner.
akostadinov
force-pushed
the
manticore
branch
from
d7764d8 to
39db158
Compare
akostadinov
force-pushed
the
manticore
branch
from
39db158 to
377448c
Compare
|
@mayorova , if you want to deploy this, the image is quay.io/3scale/searchd:manticore and in operator you can merge this into your object: spec:
system:
searchdSpec:
image: quay.io/3scale/searchd:manticore |
There was a problem hiding this comment.
I tried to build the image, I ran:
$ buildah build .
And it works until it gets stuck at 36%:
[ 36%] Building CXX object src/CMakeFiles/lmanticore.dir/sphinxsort.cpp.o
AFAIK I'm not cross-compiling, right? I'm compiling for linux/amd64.
Any idea what happens?
| - name: Sanitize Platforms | ||
| id: platforms | ||
| run: | | ||
| platforms="${{ inputs.platforms == '' && 'linux/amd64' || inputs.platforms }}" | ||
| archs="$( sed -e 's#linux/##g' <<< $platforms )" | ||
| echo "platforms=$platforms" >> $GITHUB_OUTPUT | ||
| echo "archs=$archs" >> $GITHUB_OUTPUT |
There was a problem hiding this comment.
Do we need all of this just because clang is broken in s390x. Couldn't we just use gcc?
There was a problem hiding this comment.
This part just selects platforms to build for. It's not because of s390 and will be equally needed if gcc was used. Presently build is too slow to complete for non-native platforms regardless of compiler. So we can only build for x86 in github. If we find no solution, we may need to use circleci in case we want community image to support all archs.
There was a problem hiding this comment.
btw the s390x and ppc64le clang builda were fixed upstream
Actually, it just finishes eventually. |
akostadinov
force-pushed
the
manticore
branch
from
4857601 to
ee24a4d
Compare
| SEARCHD_REPO=https://github.com/manticoresoftware/manticoresearch.git \ | ||
| CC=clang \ | ||
| CXX=clang++ \ | ||
| BUILD_PATH=/tmp/manticore_uselessly_long_path_to_prevent_rpm_build_issues \ |
There was a problem hiding this comment.
Maybe add a comment here to understand the issue?
There was a problem hiding this comment.
just fails with short path, if you have a better wording, happy to update :)
There was a problem hiding this comment.
Just some requirement of the RPM build tools. I have no idea and don't really care. Upstream they use directory with a long string of a characters as a suffix.
No description provided.