Problem:
The current approach compares the lookup time with a
predetermined time of 4.4 secs (It took 4 seconds in
RHGS 3.5.2 and keeping a buffer of 10%). But this value
is machine is specific and vairies accordingly with
machines. For e.g, below are the time taken when executed
the above tests locally
Glusterfs 49
Lookup took : 19 seconds (distributed)
Lookup took : 14 seconds (distributed-replicated)
Lookup took : 14 seconds (distributed-dispersed)
Lookup took : 14 seconds (distributed-arbiter)
Glusterfs 56.2
Lookup took : 15 seconds (distributed)
Lookup took : 10 seconds (distributed-replicated)
Lookup took : 13 seconds (distributed-dispersed)
Lookup took : 11 seconds (distributed-arbiter)
Glusterfs 59
Lookup took : 15 seconds (distributed)
Lookup took : 10 seconds (distributed-replicated)
Lookup took : 13 seconds (distributed-dispersed)
Lookup took : 10 seconds (distributed-arbiter)
Solution:
The ideal way is to uninstall gluster rpms and install an
older version and run the lookup and finally compare the
results. But this approach is not feasible with our setup.
Hence removing the time comparison check until an efficient
approach is found
[TestFix] Fix test_rebalance_with_add_brick_and_lookup
Problem: The current approach compares the lookup time with a predetermined time of 4.4 secs (It took 4 seconds in RHGS 3.5.2 and keeping a buffer of 10%). But this value is machine is specific and vairies accordingly with
machines. For e.g, below are the time taken when executed the above tests locally
Solution: The ideal way is to uninstall gluster rpms and install an older version and run the lookup and finally compare the
results. But this approach is not feasible with our setup. Hence removing the time comparison check until an efficient approach is found