Skip to contentSkip to conversation

auto-t: remove scanning wait from testConnectAutoconnect

PATCH In review James Prestwood 2 participants 5 messages May 6, 2020

lore
Share & tools
lore
#0 lore raw

PATCH 1/2 auto-t: remove scanning wait from testConnectAutoconnect

In certain cases the test was reaching this point once already finished with a scan. This caused the wait for 'scanning' to time out. At this point in the test we don't need to wait for scanning and only need to wait for scanning to finish. If scanning is already finished before validation we can simply validate at that time.

---
 autotests/testConnectAutoconnect/validation.py | 3 ---
 1 file changed, 3 deletions(-)
autotests/testConnectAutoconnect/validation.py +0 -3
index 43cbaf02..b1e3accc 100644
@@ -54,9 +54,6 @@ class TestConnectAutoConnect(unittest.TestCase):
         self.assertIsNotNone(devices)
         device = devices[0]
 
-        condition = 'obj.scanning'
-        wd.wait_for_object_condition(device, condition)
-
         condition = 'not obj.scanning'
         wd.wait_for_object_condition(device, condition)
 
Signature · 1 line
2.21.1
#1 lore raw

PATCH 2/2 auto-t: remove scanning wait from testHiddenNetworks

Same deal as with testKnownNetworks, the wait for scanning to stop has been removed since scanning may not be running at this point.

---
 autotests/testHiddenNetworks/validation.py | 2 --
 1 file changed, 2 deletions(-)
autotests/testHiddenNetworks/validation.py +0 -2
index 9c433623..551f12c7 100644
@@ -50,8 +50,6 @@ class TestConnectAutoConnect(unittest.TestCase):
         device = devices[0]
 
         if autoconnect:
-            condition = 'obj.scanning'
-            wd.wait_for_object_condition(device, condition)
             condition = 'not obj.scanning'
             wd.wait_for_object_condition(device, condition)
 
Signature · 1 line
2.21.1
#2 lore raw
In reply to James Prestwood

Hi James,

Quoted 14 lines · James Prestwood

In certain cases the test was reaching this point once already finished with a scan. This caused the wait for 'scanning' to time out. At this point in the test we don't need to wait for scanning and only need to wait for scanning to finish. If scanning is already finished before validation we can simply validate at that time.

---
  autotests/testConnectAutoconnect/validation.py | 3 ---
  1 file changed, 3 deletions(-)
autotests/testConnectAutoconnect/validation.py +0 -0
index 43cbaf02..b1e3accc 100644

autotests/testConnectAutoconnect/validation.py class TestConnectAutoConnect(unittest.TestCase): · line 59

Quoted 6 lines
autotests/testConnectAutoconnect/validation.py +0 -3
@@ -54,9 +54,6 @@ class TestConnectAutoConnect(unittest.TestCase):
          self.assertIsNotNone(devices)
          device = devices[0]
 
-        condition = 'obj.scanning'
-        wd.wait_for_object_condition(device, condition)
-

What if the scan has never started yet? Since iwd is started from Python, it looks suspicious that it goes through the scan before it hits the conditional wait a few lines down the script. Perhaps changing to wd.list_devices(1) API has introduced this issue. Would it help to increase the conditional wait time out so it goes into the second round of scans for these certain cases?

autotests/testConnectAutoconnect/validation.py class TestConnectAutoConnect(unittest.TestCase): · line 59

Quoted 3 lines
autotests/testConnectAutoconnect/validation.py +0 -0
          condition = 'not obj.scanning'
          wd.wait_for_object_condition(device, condition)
 

Best,
Tim

#3 lore raw

Hi Tim,

Quoted 21 lines · Tim A

Hi James,

Quoted 17 lines · James Prestwood

In certain cases the test was reaching this point once already finished with a scan. This caused the wait for 'scanning' to time out. At this point in the test we don't need to wait for scanning and only need to wait for scanning to finish. If scanning is already finished before validation we can simply validate at that time.

---
  autotests/testConnectAutoconnect/validation.py | 3 ---
  1 file changed, 3 deletions(-)
autotests/testConnectAutoconnect/validation.py +0 -0
index 43cbaf02..b1e3accc 100644
@@ -54,9 +54,6 @@ class TestConnectAutoConnect(unittest.TestCase):
          self.assertIsNotNone(devices)
          device = devices[0]
 
Quoted 9 lines

autotests/testConnectAutoconnect/validation.py class TestConnectAutoConnect(unittest.TestCase): · line 59

Quoted 3 lines
autotests/testConnectAutoconnect/validation.py +0 -3
-        condition = 'obj.scanning'
-        wd.wait_for_object_condition(device, condition)
-

What if the scan has never started yet? Since iwd is started from Python, it looks suspicious that it goes through the scan before it hits the conditional wait a few lines down the script. Perhaps changing to wd.list_devices(1) API has introduced this issue. Would it help to increase the conditional wait time out so it goes into the second round of scans for these certain cases?

I can try increasing the timeout, but I think relying on periodic scans for this autotest isn't the best way to go about it. I will try explicitly scanning via dbus, otherwise you are just hoping that a periodic scan was started. This timing will vary slightly between systems which is why I think I am hitting this.

Quoted 6 lines

autotests/testConnectAutoconnect/validation.py class TestConnectAutoConnect(unittest.TestCase): · line 59

Quoted 4 lines
autotests/testConnectAutoconnect/validation.py +0 -0
          condition = 'not obj.scanning'
          wd.wait_for_object_condition(device, condition)
 

Best,
Tim

#4 lore raw

Hi James,

Quoted 9 lines
Quoted 9 lines

autotests/testConnectAutoconnect/validation.py class TestConnectAutoConnect(unittest.TestCase): · line 59

Quoted 3 lines
autotests/testConnectAutoconnect/validation.py +0 -3
-        condition = 'obj.scanning'
-        wd.wait_for_object_condition(device, condition)
-

What if the scan has never started yet? Since iwd is started from Python, it looks suspicious that it goes through the scan before it hits the conditional wait a few lines down the script. Perhaps changing to wd.list_devices(1) API has introduced this issue. Would it help to increase the conditional wait time out so it goes into the second round of scans for these certain cases?

Quoted 5 lines

I can try increasing the timeout, but I think relying on periodic scans for this autotest isn't the best way to go about it. I will try explicitly scanning via dbus, otherwise you are just hoping that a periodic scan was started. This timing will vary slightly between systems which is why I think I am hitting this.

The purpose of this autotest and of the one with the hidden networks is to exercise auto-connection without an explicit call from Dbus (different code paths).