750744 - Test and deploy SUT agent 1.08

Status: RESOLVED FIXED

(Infrastructure & Operations Graveyard :: CIDuty, task, P1)

Description

[Armen [:armenzg]]

Attachment: [wip] test updating to 1.08

I am testing:
http://people.mozilla.org/~jmaher/sutagent/SUTAgentAndroid108.apk

The patch also tests a patch in review by Callek:
https://bugzilla.mozilla.org/page.cgi?id=splinter.html&bug=741478&attachment=618465

My base starting revision is default rather than eba35f2aeabd which is currently used on buildbotcustom.

Comment 1

[[e-mail] (:bhearsum)]

I think Platform Support is the right place for this...maybe Automation (General).

Comment 2

[Armen [:armenzg]]

I think there is something weird with the upgrade.
I think it makes the tegra board unusable.
I have tried upgrading tegra-018 and tegra-019.

## tegra-018.build.mtv1.mozilla.com
bash-3.2$ telnet 10.250.49.5 20701
Trying 10.250.49.5...
Connected to tegra-018.build.mtv1.mozilla.com.
Escape character is '^]'.
$>ver
SUTAgentAndroid Version 1.07
$>quit
quit
$>Connection closed by foreign host.
bash-3.2$ export SUT_NAME=tegra-019
bash-3.2$ python updateSUT.py 10.250.49.5
reconnecting socket
INFO: updateSUT.py: We're running SUTAgentAndroid Version 1.07
INFO: updateSUT.py: We're going to try to install SUTAgentAndroid Version 1.08
INFO: updateSUT.py: We're downloading the apk: http://build.mozilla.org/talos/mobile/sutAgentAndroid.1.08.apk
send cmd: updt com.mozilla.SUTAgentAndroid /mnt/sdcard/sutAgentAndroid.apk

recv'ing...
response: exit
$>
^@WARNING: updateSUT.py: We have tried to connect 1 time(s) after trying to update.
^@WARNING: updateSUT.py: We have tried to connect 2 time(s) after trying to update.
^@^@WARNING: updateSUT.py: We have tried to connect 3 time(s) after trying to update.
^@WARNING: updateSUT.py: We have tried to connect 4 time(s) after trying to update.
^@^@WARNING: updateSUT.py: We have tried to connect 5 time(s) after trying to update.
ERROR: updateSUT.py: We should have been able to get the version
ERROR: updateSUT.py: 'NoneType' object has no attribute 'sendCMD'

Comment 3

[cmtalbert]

Jmaher, can you take a look at this?

Comment 4

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: a few small tweaks to allow updateSUT to work (1.0)

Comment 5

[Armen [:armenzg]]

Comment on attachment 621982 [details] [diff] [review]
a few small tweaks to allow updateSUT to work (1.0)

The patch is empty.

Comment 6

[Joel Maher ( :jmaher ) (UTC -8)]

Attachment: patch with content in it (1.1)

Comment 7

[Armen [:armenzg]]

I applied jmaher's patch to my file on people *and* grabbed the latest sut apk.
I will test this on staging with other boards.

foopy06:build cltbld$ python updateSUT.py 10.250.49.4
reconnecting socket
INFO: updateSUT.py: We're running SUTAgentAndroid Version 1.07
INFO: updateSUT.py: We're going to try to install SUTAgentAndroid Version 1.08
INFO: updateSUT.py: We're downloading the apk: http://build.mozilla.org/talos/mobile/sutAgentAndroid.1.08.apk
send cmd: updt com.mozilla.SUTAgentAndroid /mnt/sdcard/sutAgentAndroid.apk

recv'ing...
response: exit
$>
^@reconnecting socket
INFO: updateSUT.py: We're running SUTAgentAndroid Version 1.08
INFO: updateSUT.py: We're now running SUTAgentAndroid Version 1.08

Comment 8

[Armen [:armenzg]]

Attachment: jmaher's patch plus URLException fixed

Comment 9

[Armen [:armenzg]]

Attachment: upgrade to 1.08 plus Callek's patch

jmaher: this is a small interdiff. thanks for the previous review

Comment 10

[Joel Maher ( :jmaher ) (UTC -8)]

Comment on attachment 622077 [details] [diff] [review]
upgrade to 1.08 plus Callek's patch

Review of attachment 622077 [details] [diff] [review]:
-----------------------------------------------------------------

r=me with the comments addressed below.

::: sut_tools/updateSUT.py
@@ +36,5 @@
>           data = download_apk()
>      except Exception, e:
>           print "ERROR: updateSUT.py: We have failed to retrieve the SUT Agent. %s" % str(e)
>           return RETCODE_APK_DL_FAILED
> +    remoteAPK = apkfilename.split('/')[-1]

we can just use apkfilename since it is split out from the foopydir above, here and in 2 places below.

@@ +49,5 @@
>      ver = None
>      tries = 0
>      while tries < 5:
>          try:
> +            dm = connect(os.getenv('SUT_IP'), sleep=90)

connect is a boolean, you could put 1 or 31415926 in and it would act the same.  Making this a boolean (i.e. True) prevents some confusion in the future.

Comment 11

[Justin Wood (:Callek)]

Comment on attachment 622077 [details] [diff] [review]
upgrade to 1.08 plus Callek's patch

From what I can tell with my patch applied, the only change needed is the actual version at top of file. But that said, adding comments inline

>-def doUpdate(dm):
>+def doUpdate(dm, deviceIP):

With the change below (to use os.env) we don't use this anywhere

>-    dm.sendCMD(['push /mnt/sdcard/%s %s\r\n' % (apkfilename, str(len(data))), data], newline=False)
>+    remoteAPK = apkfilename.split('/')[-1]
>+    dm.sendCMD(['push /mnt/sdcard/%s %s\r\n' % (remoteAPK, str(len(data))), data], newline=False)
>     dm.debug = 5
>-    dm.sendCMD(['updt com.mozilla.SUTAgentAndroid /mnt/sdcard/%s' % apkfilename])
>+    dm.sendCMD(['updt com.mozilla.SUTAgentAndroid /mnt/sdcard/%s' % remoteAPK])

These changes are no longer necessary, since apkfilename is not a filepath anymore.

>-            dm = connect(deviceIP, sleep=90)
>+            dm = connect(os.getenv('SUT_IP'), sleep=90)

using sleep as a boolean is fine by me as joel suggested, but I chose not to change it in my patch -- up to you.

>-        return doUpdate(dm)
>+        return doUpdate(dm, deviceIP)

see above

r+ given what I said above.

Comment 12

[Armen [:armenzg]]

Attachment: [tools] update SUT to 1.08

Comment 13

[Armen [:armenzg]]

Attachment: [buildbotcustom] use newer updateSUT.py

Comment 14

[Justin Wood (:Callek)]

Comment on attachment 622821 [details] [diff] [review]
[tools] update SUT to 1.08

Review of attachment 622821 [details] [diff] [review]:
-----------------------------------------------------------------

would rather have had the non-ver changes in a separate bug than tied here (since its harder to go back in time and find reasons for them) but not worth the trouble since they are in this patch and quite minor

Comment 15

[Justin Wood (:Callek)]

Comment on attachment 622823 [details] [diff] [review]
[buildbotcustom] use newer updateSUT.py

Need to add in the sut_lib get here as well, like you have locally for your testing

Comment 16

[Armen [:armenzg]]

Attachment: [buildbotcustom] use newer updateSUT.py

Comment 17

[Mike Taylor [:bear]]

Comment on attachment 623657 [details] [diff] [review]
[buildbotcustom] use newer updateSUT.py

did you mean to have a real revision number in the wget for updateSUT.py?

Comment 18

[Armen [:armenzg]]

Comment on attachment 622821 [details] [diff] [review]
[tools] update SUT to 1.08

http://hg.mozilla.org/build/tools/rev/16fc4f354b44

Comment 19

[Armen [:armenzg]]

Attachment: [buildbotcustom] use newer updateSUT.py

with revision.

Comment 20

[Mike Taylor [:bear]]

Comment on attachment 623669 [details] [diff] [review]
[buildbotcustom] use newer updateSUT.py

r+'ing only to get this out of the ateam blocking status

Comment 21

[Armen [:armenzg]]

Comment on attachment 623669 [details] [diff] [review]
[buildbotcustom] use newer updateSUT.py

http://hg.mozilla.org/build/buildbotcustom/rev/36b6b4b6d9d4

Comment 22

[Armen [:armenzg]]

All foopies have been updated to:
16fc4f354b44

I will now be watching that everything stays put.

Comment 23

[Armen [:armenzg]]

A change from this bug landed on the default branches of either buildbot-configs or buildbotcustom.
Such change has been merged as of ~8:30 AM PDT into the production branches and the buildbot masters have been reconfigured.
This if FYI and a note has been added to the maintenance page:
https://wiki.mozilla.org/ReleaseEngineering:Maintenance#Reconfigs_.2F_Deployments

Comment 24

[Justin Wood (:Callek)]

You forgot to update foopy24 tools along with the others, and explains why foopy24 was exhibiting lots of issues per philor today.

I just updated it, and will check the other foopies shortly, for sanity.

Comment 25

[Armen [:armenzg]]

(In reply to Justin Wood (:Callek) from comment #24)
> You forgot to update foopy24 tools along with the others, and explains why
> foopy24 was exhibiting lots of issues per philor today.
> 

I rock...not :(

Now I know that 24 is the last one of the foopies. All the other ones were updated so no need to check.

Thanks Callek. sorry.