GNOME Bugzilla – Bug 659491
Hung when accepting calendar invite
Last modified: 2012-07-03 07:13:32 UTC
Most calendar invites crash my current 3.1.91 evolution (Ubuntu 11.10 B1). Downstream I have captured two scenarios: 1) Declining a calendar invite resulting in an e-cal-factory crash and auto-tracing e-calendar-factory crashed with SIGSEGV in _dbus_watch_invalidate() https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/853986 2) Trying to accept a calendar invite resulting in a hang that I captured at forced quit with gdb https://bugs.launchpad.net/ubuntu/+source/evolution/+bug/850218 I'm not entirely sure these are related, but my Google Calendars are not really working at all from an accept/decline perspective.
Can you please paste traces here from downstream bug, tia ?
for the point 2 here the BT (gdb) backtrace full
+ Trace 228495
Could you open a separate bug report for 1), please? Please mention there what calendar type is used here (On This Computer/CalDAV/...) For me the application hangs due to a sync call through the DBus to e-calendar-factory, and it recovers after some time (it was a long time, anyway).
#1 moved here https://bugzilla.gnome.org/show_bug.cgi?id=659756
This one is still top of my list in 3.2.0. When I preview an email to "Accept" I don't even see the "Send Reply" checkbox if the event is not already in my calendar. Also some multipart appointments show their .ics attachment and others do not.
It would be helpful to provide test messages, because mine works fine. (Of course with sanitized private information like email addresses and so on.)
Created attachment 197905 [details] Mbox Attachment Attached is an mbox file with an email that contains an ICS calendar event. The resulting crash appears retraced as https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/863590 I am not convinced this is actually different than https://bugzilla.gnome.org/show_bug.cgi?id=659756 In this case, it is an email from one of my IMAP accounts to another one. I created it in the browser at Google.com/calendar. Both the meeting guest and organizer calendars are configured in my Evolution, so it showed as "Found in Calendar" even before I accepted it, however the issue occurs in a similar manner from external calendars as well. I cannot accept any invite from any client as far as I know (Google, Outlook, etc).
OK, I think I misunderstood this then. To clarify, you can see the buttons for Accept/Decline/... when you select the message and selecting it doesn't freeze the evolution, the issue becomes only when you click on one of those Accept/Decline/Tentative buttons, which can result in a crash on the e-calendar-factory side, and an infinite hang on evolution's side (because it's waiting on a reply from the factory, which will never happen)? I suppose you've destination calendar is a Gnome-Online-Account Google calendar, am I right? I still do not have updated my system, but I should get to it soon.
Milan, your clarification is accurate.
And it does not happen with local calendar
There is no output from the calendar factory when it dies.
(In reply to comment #11) > There is no output from the calendar factory when it dies. Did it close itself, with a "Bye" message, or crashed? Running it under gdb may tell you where it crashed, if it did.
Oh it crashes, just no console output. It is retraced in the downstream link as it caught immediately by apport. https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/863590
I noticed there are a number of duplicate reports being caught by apport, so hopefully this is easily reproduced.
(In reply to comment #13) > Oh it crashes, just no console output. It is retraced in the downstream link > as it caught immediately by apport. > > https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/863590 Reported in Gnome as bug #659756, though I'm not seeing it on Fedora 15 myself.
Seen this in GNOME 3.2 on Fedora 16.
I can't seem to reproduce this now with current 3.2.1-0ubuntu1 on my Ubuntu 11.10 instance. This clears the path for an upgrade of my daily driver.
After a few more days I believe this is resolved in 3.2.1-0ubuntu1.
may be, this bug can be closed while keeping bug 659756 open
Probably. Both seem to be fixed for Ian with 3.2.1, as stated in comment #18 and bug #659756 comment #17.
Well, I had one potential re-occurrence almost two weeks ago, and then a second potential yesterday. In both cases I was doing too many things to be sure it was the invite. I let it trace and log downstream anyways in both previous cases: https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/890329 https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/885796 Today, I only executed one action, to accept an invite, only to find the e-cal factory crash moments later when I glanced back to my screen. This confirmed that this issue is still around, just on every invite as it was previously in my case. I logged this last crash too in case it is helpful;it should become public downstream in a bit. https://bugs.launchpad.net/ubuntu/+source/evolution-data-server/+bug/890787 In this last case, the issue seems 100% repeatable, occurring again after force-shutdown and reload on the same event. I have saved the email as an mbox so I can test some additional scenarios and turn on debugging.
edit my last %s/just on every invite/just not on every invite/g
The first two are still private, unfortunately. The last is public already, and they found a duplicate. I noticed a similar one as well, it seems like either a memory corruption or a dbus issue and its thread unsafety (the dbus-glib's not GDBus'). Though I cannot tell for sure.
*** Bug 663420 has been marked as a duplicate of this bug. ***
Created attachment 202120 [details] [review] evo patch for evolution; This removes almost all synchronous calls to ECalClient from itip-formatter, which were causing this hung (the last one does not do any remote I/O operation, because the property is cached on ECalClient open). The itip-formatter shows a progress information about "updating event on the server", and user should wait till it's done, because moving away from the message cancels any pending requests, like the event update. I added couple new strings, thus not possible to push to 3.2.x as is.
Created commit 8664187 in evo master (3.3.3+)
Hmm, so is there a way to create a patch that could be used for the downstream issue, and not committed, noting that it appears to affect quite a few people. https://bugs.launchpad.net/evolution-data-server/+bug/847578
Do you mind of new translation strings? Maybe I can try to provide a 3.2.x patch. Also note that more changes might be needed, for example, I recently fixed similar issue in the WebCal backend, which was crashing due to operation cancelling too.
Hmm, well they just released Ubuntu 12.04 Beta1 with 3.2.3, so this continues to plague my Ubuntu experience. Accepting calendar invites with ICS phone widgets is killing me slowly.
All fixed now downstream in Ubuntu 12.04. No longer an issue for me.
*** Bug 663223 has been marked as a duplicate of this bug. ***