Resolves #117 Attempts to retry failed mounts of bricks

[gonoph]

Resolves Issue #117

After the update in commit 2e2e284, gluster-setup.sh no longer attempts to mount failed bricks.

• checks for the return codes 1, 2, and 4 which is fatal
• logs successful mount
• if unsuccessful, allows script to continue to attempt the retry logic below.

According to the man page RETURN CODE 1 is incorrect permissions or invocation (bad parameters). The previous commit attempted to only test for that error, which if continuing in the script logic allows the glusterd pod/container to attempt to identify the failed bricks, and try various tricks to mount them.

This change tests for RETURN CODE 1, 2, and 4 - which should cover most fatal errors, and still allow the script to attempt to mount failed bricks using the logic below that test.

Also, the test at the end:

        if [ "$((cat $GLUSTERFS_LOG_CONT_DIR/failed_bricks | wc -l) )" -gt 1 ]

works better than the wc -l $GLUSTERFS_LOG_CONT_DIR/failed_bricks it replaces, as the wc command appends the filename, which breaks the number test -gt. Therefore, the cat | wc syntax works better as wc does not print the filename when used on stdin, and thus it does not break the -gt test.

[nixpanic]

Can you have a look at #114? It should address the same problem.