On first run after the app version changes ping times are worse

NEW
Unassigned

Status

()

defect
7 years ago
7 years ago

People

(Reporter: ashughes, Unassigned)

Tracking

unspecified
x86_64
Linux
Points:
---

Firefox Tracking Flags

(firefox15 affected, firefox16 wontfix, firefox17-, firefox18-)

Details

(Whiteboard: QARegressExclude)

Attachments

(2 attachments)

Discovered through investigating bug 791099.

1. Install Firefox 14.0.1
2. Load speedtest.net and run a few tests noting the ping times (~75ms for me)
3. Update to Firefox 15.0.1 through the About dialog and restart
4. Load speedtest.net and run a few tests noting the ping times (~125ms for me)

I realize this is not a scientific test but it appears Software update regresses ping times considerably. If I do a clean install of Firefox 15.0.1 this regression is not evident.
Could you add the app.update.log true preference, run through your steps, and check to see if there are any AUS:SVC error console messages? This will determine if it is due to the ping or something else.
Does this happen on a second run of the speedtest during the same session? Does it happen again after a restart?
I'm confused. This log shows a manual check and that there is a download of an update happening. Are you asking that if we are downloading an update (manual or in the background) that we shouldn't affect network speeds?

Are the speedtests affected when you aren't downloading an update?
Attachment #665600 - Attachment description: Copy of AUS log → AUS log before restart
(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #5)
> Created attachment 665604 [details]
> AUS log after restart
AUS isn't touching the network in this log so if it is happening here it isn't the cause. Might be due to the app cache being rebuilt after an update but that isn't under app update's control.
(In reply to Robert Strong [:rstrong] (do not email) from comment #2)
> Does this happen on a second run of the speedtest during the same session?
> Does it happen again after a restart?

Hopefully these results answer both of your questions:

1. Install Firefox 14.0.1
2. Speedtest.net test 1: ping 65ms
3. Speedtest.net test 2: ping 55ms
4. Speedtest.net test 3: ping 55ms
Averages out to 58ms ping before update

5. Update Firefox via the About dialog and restart
6. Speedtest.net test 1: ping 125ms
7. Speedtest.net test 2: ping 125ms
8. Speedtest.net test 3: ping 125ms
Averages out to 125ms after update

9. Restart Firefox
10. Speedtest.net test 1: ping 55ms
11. Speedtest.net test 2: ping 65ms
12. Speedtest.net test 3: ping 65ms
Averages out to 62ms after restart
So, 5, 6, 7, 8 are from "AUS log after restart"?

If so, this appears to be due to application code after an update. Possibly due to the app cache being invalidated after a change or something else. There isn't a ping or a download happening in this log. There is the cleanup update of the update that was just applied which is trivial and doesn't touch the network, etc.
(In reply to Robert Strong [:rstrong] (do not email) from comment #4)
> I'm confused. This log shows a manual check and that there is a download of
> an update happening. Are you asking that if we are downloading an update
> (manual or in the background) that we shouldn't affect network speeds?

Nope, I'm not asking anything really -- I'm just reporting the ping time results at bajaj's request. I would expect anything using network to affect total bandwidth.

> Are the speedtests affected when you aren't downloading an update?
1. Install Firefox 14.0.1 and run a speedtest
> 55ms ping
2. Check for update and run a speedtest while it downloads
> 225ms ping
3. Run a speedtest before restarting to apply the update
> 75ms ping
4. Run a speedtest after restarting and applying the update
> 119ms ping
5. Restart Firefox once more and run a speedtest
> 55ms ping

2. Speedtest during update:
3. Speedtest after update:
4. Speedtest after restart:
(In reply to Anthony Hughes, Mozilla QA (:ashughes) from comment #9)
> 2. Speedtest during update:
> 3. Speedtest after update:
> 4. Speedtest after restart:

Sorry ignore this junk ^
(In reply to Robert Strong [:rstrong] (do not email) from comment #8)
> So, 5, 6, 7, 8 are from "AUS log after restart"?

Correct.

> If so, this appears to be due to application code after an update. Possibly
> due to the app cache being invalidated after a change or something else.
> There isn't a ping or a download happening in this log. There is the cleanup
> update of the update that was just applied which is trivial and doesn't
> touch the network, etc.
Can you also check by doing the following with the same profile:
Install Firefox 14.0.1
Start Firefox
Run the speedtest
Install Firefox 15
Start Firefox
Run the speedtest

and report back the numbers.

If this shows the anomaly I should be able to put together a "fake" update so the actions that happen in app update after it is applied runs so we can see if that has any affect.
(In reply to Robert Strong [:rstrong] (do not email) from comment #12)
> Can you also check by doing the following with the same profile:
> Install Firefox 14.0.1
> Start Firefox
> Run the speedtest
> Install Firefox 15
> Start Firefox
> Run the speedtest
> 
> and report back the numbers.
My concern is that if there was a second restart it would end up in the 9, 10, 11, 12 section in comment #7
Speedtest after installing and starting Firefox 14.0.1
> 65ms ping
Speedtest after pave-over install and starting Firefox 15.0.1
> 125ms ping
That shows that this isn't caused by app update and is likely caused by the application cache being invalidated / rebuilt or some other code.

Moving to Toolkit -> General
Component: Application Update → General
Summary: Application Update appears to make for worse ping times → On first run after the app version changes ping times are worse
Anthony let us know that this is resolved after application restart, which makes it not critical for release.
Whiteboard: QARegressExclude
You need to log in before you can comment on or make changes to this bug.