test: EC Small parallel writings

This test checks issues found during development of the new write
approach. The idea is to test the client behavior when many small (1
byte long) random writes are being handled in parallel (10 mounts).

tests/test_suites/ShortSystemTests/test_ec_small_parallel_writing.sh

@@ -0,0 +1,42 @@
+if is_windows_system; then
+	timeout_set 5 minutes
+else
+	timeout_set 3 minutes
+fi
+
+
+CHUNKSERVERS=4 \
+	MOUNTS=10 \
+	MOUNT_EXTRA_CONFIG="sfscachemode=NEVER" \
+	MASTER_CUSTOM_GOALS="10 ec_3_1: \$ec(3,1)" \
+	USE_RAMDISK=YES \
+	setup_local_empty_saunafs info
+
+# Create an empty file on the SaunaFS filesystem with ec_3_1 goal
+file="${info[mount0]}/file"
+touch "$file"
+saunafs setgoal ec_3_1 "$file"
+
+# Create a temporary file with 25 megabytes of generated data
+tmpf=$RAMDISK_DIR/tmpf
+FILE_SIZE=64K file-generate "$tmpf"
+
+# Run 10 tasks, each of them will copy every tenth kilobyte from the temporary file to the file
+# on SaunaFS using 5 concurrent writers and a dedicated mountpoint
+for i in {0..9}; do
+	(
+		file="${info[mount${i}]}/file"
+		seq $i 10 $((64*1024-1)) | shuf | expect_success xargs -P5 -IXX \
+				dd if="$tmpf" of="$file" bs=1 count=1 seek=XX skip=XX conv=notrunc 2>/dev/null
+	) &
+done
+wait
+
+# Validate the result
+MESSAGE="Data is corrupted after writing" expect_success file-validate "$file"
+
+# Validate the parity part
+csid=$(find_first_chunkserver_with_chunks_matching 'chunk_ec2_1_of_3*')
+saunafs_chunkserver_daemon $csid stop
+
+MESSAGE="Parity is corrupted after writing" expect_success file-validate "$file"