• Similarly to coreutils we need recent util-linux because if e.g. a testnode has
    old version of it its lsblk does not work to what neotest expects, e.g.:
    
    	lsblk: unknown column: REV,SIZE
    by Kirill Smelkov
     
    Browse Files
  • For example we need realpath and if base os provides only older coreutils there
    is no such utility and neotest fails:
    
    /srv/slapgrid/slappart8/srv/testnode/bct/soft/55158b6a5b027aa40954c3f974d5b3e9/go.work/src/lab.nexedi.com/kirr/neo/go/neo/t/neotest: line 589: realpath: command not found
    
    Fix it by building & hooking into $PATH coreutils by ourselves.
    by Kirill Smelkov
     
    Browse Files
  • …exedi testing infrastructure
    
    Commit 31fe231f (neotest: Draft software-release to run NEO/go & friends
    tests/benchmarks under webrunner) added "neotest" software release, but
    there it added only software supplying - not instantiation - this way
    allowing only manual play with neotest.
    
    Recently @rafael approached me saying that one of my webrunner is red on
    monitoring because "neotest" software inside fails to instantiate, and
    suggested to create a simple instance with bin/runTestSuite inside, so
    that this way neotest could be added to a testnode and run on regular
    basis by our testing infrastructure.
    
    So here it goes - simple instance with only bin/runTestSuite which tries
    to adhere to testnode protocol.
    
    For the actual runTestSuite program I followed the same approach as for
    bin/neotest itself - the main program is not processed via any templates
    and is used as it was pristinely downloaded, and all buildout
    integration is done via first sourcing $SR/neotest-env.sh .
    So bin/runTestSuite inside instance follows this approach.
    
    erp5.util is updated from 0.4.49 to latest 0.4.50 because there were
    backward-incompatible changes in 0.4.50 release (2 @luke: which btw will
    render neoppod and build-rina runTestSuite(s) broken when they try to
    upgrade) and for me it would be a waste of time to later go through
    updating for 0.4.49 -> 0.4.50 API change, so I used the latest release
    right from start.
    
    /cc @Tyagov
    /see also kirr/neo@526491d3
    by Kirill Smelkov
     
    Browse Files
  • by Kirill Smelkov
     
    Browse Files
  • lsblk is needed for neotest (see kirr/slapos@7cc37e3c).
    However lsblk is built only when the following dependecies are enabled:
    
    	- libblkid
    	- libmount
    	- libsmartcols
    
    	https://github.com/karelzak/util-linux/blob/v2.31-286-g8e782a839/configure.ac#L1483
    
    we already have libblkid and libmount enabled, but libsmartcols was missing.
    
    Enable it.
    by Kirill Smelkov
     
    Browse Files




  • For example one could want to build the software with race detector,
    e.g.
    
    	[gowork]
    	...
    	buildflags = -race
    
    or if you want to build a plugin for fluent-bit
    
    	[gowork]
    	...
    	install	   = your.package.import.path
    	buildflags = -buildmode=c-shared
    
    and then your package will be installed as shared library into
    
    	${gowork:directory}/pkg/`go env GOOS`_`go env GOARCH`/<your.package.import.path>.a
    
    One could pass that location explicitly to `fluentd-bit -e ...` or
    symlink to it from some .so from global directory of fluent-bit plugins.
    
    Maybe it is not very good to use `go install` for building plugins
    (instead of `go build -o ...`) but we anyway need to have ability to
    specify buildflags as a generic feature, hence this patch.
    
    /reviewed-on !274
    by Kirill Smelkov
     
    Browse Files
  • @alain.takoudjou , @rafael  this fixes broken webrunner tests.
    
    Passing one: https://nexedi.erp5.net/test_result_module/20180126-5FF058E7/2
    
    /reviewed-on !277
    by Ivan Tyagov
     
    Browse Files



  • If we don't specify MaxRequestWorkers, it seems like Apache 2.4
    errors with:
    AH03490: scoreboard is full, not at MaxRequestWorkers. Increase
    ServerLimit.
    Maybe because MaxRequestWorkers then defaults to a too high value
    compared to ServerLimit and ThreadsPerChild.
    
    /reviewed-on !263
    by Aurélien Vermylen
     
    Browse Files