Commit 68740678 authored by Michal Berger's avatar Michal Berger Committed by Tomasz Zawadzki
Browse files

test/blobfs: Allocate enough memory for db_bench tests 



Currently, BlobFS-nightly-autotest jobs are failing due to this test
not having enough memory around. Change that by allocating some
overhead depending on what CACHE_SIZE was set to.

Change-Id: I9b122189d39a7a73ab8521f699324516d09ae1c3
Signed-off-by: default avatarMichal Berger <michalx.berger@intel.com>
Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/3941


Community-CI: Mellanox Build Bot
Tested-by: default avatarSPDK CI Jenkins <sys_sgci@intel.com>
Reviewed-by: default avatarBen Walker <benjamin.walker@intel.com>
Reviewed-by: default avatarTomasz Zawadzki <tomasz.zawadzki@intel.com>
parent 150339cc

test/blobfs/rocksdb/rocksdb.sh

+9 −0
Original line number Diff line number Diff line
@@ -91,6 +91,15 @@ else 
	DURATION=20

	NUM_KEYS=20000000

fi

# Make sure that there's enough memory available for the mempool. Unfortunately,

# db_bench doesn't seem to allocate memory from all numa nodes since all of it

# comes exclusively from node0. With that in mind, try to allocate CACHE_SIZE

# + some_overhead (1G) of pages but only on node0 to make sure that we end up

# with the right amount not allowing setup.sh to split it by using the global

# nr_hugepages setting. Instead of bypassing it completely, we use it to also

# get the right size of hugepages.

HUGEMEM=$((CACHE_SIZE + 1024)) HUGENODE=0 \

	"$rootdir/scripts/setup.sh"



cd $RESULTS_DIR

cp $testdir/common_flags.txt insert_flags.txt