glusto-tests.git/tests/functional/afr, branch master

[Test]Test split-brain with hard link and brick down scenario

2021-02-10T06:38:20+00:00

Change-Id: Ib58a45522fc57b5a55207d03b297060b29ab27cf
Signed-off-by: Manisha Saini

[Testfix] Fix I/O logic and change lookup command

2021-02-02T09:47:23+00:00

Problem:
1. I/O command has %s instead of %d for for int.
2. Lookup logic doesn't tigger client heal.

Fix:
Changing to %d and adding a cmd which automatically
triggers heal.

Change-Id: Ibc8a1817894ef755b13c3fee21218adce3ed9c77
Signed-off-by: kshithijiyer

[Test] Add test to test split brain with node reboot

2021-01-25T05:38:08+00:00

Change-Id: Ic5258b83b92f503c1ee50368668bd7e1244ac822
Signed-off-by: Manisha Saini

[Test] Add test to check default granular entry heal

2021-01-25T05:31:48+00:00

Test case:
1. Create a cluster.
2. Create volume start it and mount it.
3. Check if cluster.granular-entry-heal is ON by default or not.
4. Check /var/lib/glusterd//info for
   cluster.granular-entry-heal=on.
5. Check if option granular-entry-heal is present in the
   volume graph or not.
6. Kill one or two bricks of the volume depending on volume type.
7. Create all types of files on the volume like text files, hidden
   files, link files, dirs, char device, block device and so on.
8. Bring back the killed brick by restarting the volume.
9. Wait for heal to complete.
10. Check arequal-checksum of all the bricks and see if it's proper or
    not.

Refernce BZ: #1890506
Change-Id: Ic264600e8d1e29c78e40ab7f93709a31ba2b883c
Signed-off-by: kshithijiyer

[Testfix] Fix I/O logic and use text for getfattr

2021-01-22T09:28:34+00:00

Problem:
1. The code uses both clients to create files
   with the same names, causing I/O failures
   in one set of clients
2. Code tries to match hex value of
   replica.split-brain-status to text string.

Solution:
Fix I/O logic to use only one client and getfattr
in text for proper comparision.

Change-Id: Ia0786a018973a23835cd2fecd57db92aa860ddce
Signed-off-by: kshithijiyer

[Test] Add test to check self heal with expand volume

2021-01-21T06:21:09+00:00

Test case:
1. Created a 2X3 volume.
2. Mount the volume using FUSE and give 777 permissions to the mount.
3. Added a new user.
4. Login as new user and created 100 files from the new user:
   for i in {1..100};do dd if=/dev/urandom of=$i bs=1024 count=1;done
5. Kill a brick which is part of the volume.
6. On the mount, login as root user and create 1000 files:
   for i in {1..1000};do dd if=/dev/urandom of=f$i bs=10M count=1;done
7. On the mount, login as new user, and copy existing data to
   the mount.
8. Start volume using force.
9. While heal is in progress, add-brick and start rebalance.
10. Wait for rebalance and heal to complete.
11. Check for MSGID: 108008 errors in rebalance logs.

Refernce BZ: #1821599
Change-Id: I0782d4b6e44782fd612d4f2ced248c3737132855
Signed-off-by: kshithijiyer

[Test+Lib] Add tests to check self heal

2021-01-20T05:59:20+00:00

Test scenarios added:
1. Test to check entry self heal.
2. Test to check meta data self heal.
3. Test self heal when files are removed and dirs
   created with the same name.

Additional libraries added:
1. group_del(): Deletes groups created
2. enable_granular_heal(): Enables granular heal
3. disable_granular_heal(): Disables granular heal

Change-Id: Iffa9a100fddaecae99c384afe3aaeaf13dd37e0d
Signed-off-by: kshithijiyer

[Test+Libfix] Add test to add brick followed by remove brick

2021-01-18T05:07:50+00:00

Test case:
1. Create a volume, start it and mount it to a client.
2. Start I/O on volume.
3. Add brick and trigger rebalance, wait for rebalance to complete.
   (The volume which was 1x3 should now be 2x3)
4. Add brick and trigger rebalance, wait for rebalance to complete.
   (The volume which was 2x3 should now be 3x3)
5. Remove brick from volume such that it becomes a 2x3.
6. Remove brick from volume such that it becomes a 1x3.
7. Wait for I/O to complete and check for any input/output errors in
   the logs of both the I/O and rebalance logs.

Additional library fix:
Adding `return True` at the end of is_layout_complete()
to return True if no issues found in layout.

Refernce BZ: #1726673
Change-Id: Ifd0360f948b334bfcf341c1015a731274acdb2bf
Signed-off-by: kshithijiyer

[Testfix] Fix setup bug in test_no_glustershd_with_distribute

2021-01-13T05:33:20+00:00

Problem:
In test_no_glustershd_with_distribute, we are trying to
setup all volume types at once where it fails on setup
in CI as we don't have sufficiant bricks.

Solution:
Enable brick sharing in setup_volume() by setting
multi_vol to True.

Change-Id: I2129e3059fd156138d0a874d6aa6904f3cb0cb9b
Signed-off-by: kshithijiyer

[Test] Add tests to check heal with hard and soft links

2021-01-04T06:13:36+00:00

Test Scenarios:
---------------
1. Test heal of hard links through default heal
2. Test heal of soft links through default heal

CentOS-CI failing due to issue:
https://github.com/gluster/glusterfs/issues/1954

Change-Id: I9fd7695de6271581fed7f38ba41bda8634ee0f28
Signed-off-by: kshithijiyer