Closed Bug 1792048 Opened 2 years ago Closed 2 years ago

Thunderbird doesn't reliably synchronize with Google Calendar since 102.2

Categories

(Calendar :: Provider: CalDAV, defect)

Thunderbird 102
defect

Tracking

(Not tracked)

RESOLVED DUPLICATE of bug 1789424

People

(Reporter: ewl+mozilla, Unassigned)

References

()

Details

Attachments

(3 files)

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:104.0) Gecko/20100101 Firefox/104.0

Steps to reproduce:

  1. yum upgrade from thunderbird-91.12.0-1.fc36.x86_64 to 102.2.1 (on Fedora 36)
  2. reboot and open Thunderbird

Actual results:

Since the upgrade, my 8 calendars only synchronize very unreliably, sometimes to the point that my calendar is empty even though it should be fully packed. Calendars get marked with a caution sign, sometimes they even get disabled and I need to re-enable, so this is a complete mess and I can't currently rely on Thunderbird to do calendaring, which is a pity because I appreciate the integration with my favourite mailer.

Expected results:

I can trust Thunderbird calendar to show all my appointments and changes reliably.
It used to work more or less correctly (there were glitches already but nothing that bad). I have provided already some more information at https://support.mozilla.org/en-US/questions/1389156 and someone claimed to have the same issue, but let me know which information you need and I'll provide it.

The 8 calendars are made of 6 professional calendars with oauth connection to Google and 2 private ones with normal user/password login. They seem to be all equally impacted but it's rather random.

I (and others) have seen this quite frequently. I don't see any other bugs on file to duplicate this into.

Status: UNCONFIRMED → NEW
Component: Untriaged → Provider: CalDAV
Ever confirmed: true
Product: Thunderbird → Calendar
See Also: → 1791771

Like Patrick said there's been a few weird calendar things going on lately, and they all seem to involve Google. I'm wondering if something is going on with their server.

Reading your support.m.o report, it is strange that you're getting network errors immediately after creating the calendar. Most reports I've seen come from calendars that have been set up for some time already.

Could you open the network tab of the developer toolbox, reproduce the problem, then post the contents of the network request/response?

Flags: needinfo?(ewl+mozilla)

So this morning, around 9, everything was working fine, after I had upgraded from thunderbird-102.3.0-1.fc36.x86_64 to thunderbird-102.2.1-1.fc36.x86_64, so I thought that it might have solved the issue.
Then it went belly up. The two screenshots I just added show how things evolved as I was having a call, and didn't touch at all my mailer. The issue just happened which makes it difficult to "reproduce" and track. Do you have a recommendation on how to do this the best way for you to analyze?

Here my actions, with result and corresponding archive file, cleaned-up before each action:

  • 'messenger_Archive [22-09-26 10-58-04].har' -> synchronize all calendars, no apparent change
  • 'messenger_Archive [22-09-26 10-59-41].har' -> enable disabled calendar, still warning mode
  • 'messenger_Archive [22-09-26 11-00-29].har' -> enable another disabled calendar, warning gone
  • 'messenger_Archive [22-09-26 11-01-51].har' -> enable yet another disabled calendar, stayed disabled
  • 'messenger_Archive [22-09-26 11-03-00].har' -> try again, succeed this time
  • 'messenger_Archive [22-09-26 11-06-29].har' -> synchronize all calendars, 2 calendars disabled, all with warning signs (calendar is completely empty)

Let me know if I can do more/better.

Flags: needinfo?(ewl+mozilla)

Ups, before I upload, i realize that the files are apparently full of secrets, tokens and IDs, is there a (simple) way to anonymize/sanitize before uploading?

Flags: needinfo?(vseerror)
Flags: needinfo?(vseerror)

I'm most interested in the method, url and post data parts of the request section, and the status and content parts of the response section. I probably don't need the headers. Doing a find/replace on calendar identifiers (your email address or anything ending with %40group.calendar.google.com) should make all other tokens useless (they probably were anyway).

Alternatively you could email me the HAR files at this address and they won't go any further.

I've edited out the "crispy" parts, hopefully I didn't oversee anything important (I wasn't sure what "this address" means and I'm eager to get this issue fixed quickly)

For reference, what I edited out (Vim commands):

     33  %s/<company>/acme/g
     34  %s/<user>/myuser/g
     35  %s/<privateuser>/privateme/g
     36  %s/acme.com_[^@]*%40group.calenda..._somecalendarID%40group.calendar/g
     37  %s/acme.com_[^@ ]*@group.calendar...om_somecalendarID@group.calendar/g
     39  %s/<client_secret>/someclientsecret/g
     40  %s/<client_id>/someclientid/g
     41  %s/family[^@ ]*%40/familymyfamily%40/g
     43  %s/"text": "\(<[^>]*>\).*\(<[^<>]*>\\n\)"/"text": "\1editedtext\2"/g
     44  %s/"text": "\(<[^>]*>\).*\(<[^<>]*>\)"/"text": "\1editedtext\2"/g
     48  %s/family[^@ ]*@/familymyfamily@/g

Thanks. Looks like this might be the same problem as bug 1789424. If you remove one of the affected calendars and re-add it, does it work properly, at least until this happens again?

Flags: needinfo?(ewl+mozilla)

I've already tried (see my first post, as linked) and it didn't help at that time. As another remark, the phenomen seems to be in phases: right now all looks good, but two hours ago, my calendar was almost empty, nothing being synced.

Uups, exactly as I write those lines, my calendar went almost empty within 5 (or so) seconds, and at the same time there was a surge in the fan's noise. It took me more time to write this sentence than for this to happen. Forcing a sync brings back only few entries but the main calendar is now "momentarily not available". Second sync doesn't bring any change.

Flags: needinfo?(ewl+mozilla)

For what it's worth, I tried to keep track of the state of my calendar and it's simply going up and down. Of course I haven't been looking the whole time so I have surely missed some of the waves, but it doesn't seem to be tied to any action I take, or time of the day (Google overloaded or so)

2022-09-29 9:00 empty
2022-09-29 11:00 full, then empty
2022-09-29 12:09 empty
2022-09-29 14:38 almost all calendars disabled, reenabling them, few entries appear
2022-09-29 14:44 full calendar, back to normal
2022-09-29 15:03 calendar almost empty again, 2 calendars are available, only one shows entries, others are marked unavailable
2022-09-29 18:00 5 calendars disabled, re-enabled all of them, calendar fills up
2022-09-29 18:02 calendar back to normal
Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: