GNOME Bugzilla – Bug 328865
calendar and contacts lose Exchange connectivity, mail continues to function
Last modified: 2007-05-16 20:19:56 UTC
Please describe the problem: I start up evolution, enter my OWA URL and all other setup information, and evo works fine for a period of up to a week. Then at some point the contacts and calendaring functionality lose their ability to get information from Exchange; the first indication that I get that this has happened is that address autocompletion stops working in the mail composition window. After this has happened, the following behavior occurs: When I select the Contacts area and click on the GAL, I get an error popup "Error loading addressbook. We were unable to open this addressbook. This either means you have entered an incorrect URI, or the server is unreachable." At the same time, the debug output (I have E2K_DEBUG and CAMEL_DEBUG set) on my terminal reads: calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) (evolution:22543): eab-widgets-WARNING **: EABView at present does not support multiple writes on the "source" property. When I select the Calendar view, the checkbox on my remote calendar unchecks itself, and I get debug output of: (evolution:22543): eab-widgets-CRITICAL **: eab_view_discard_menus: assertion `view->view_instance' failed If I try to re-check the remote calendar, it unchecks immediately with: calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) If I then go back to the Mail view and open a new message, each keystroke (except the first two) in the Address or CC fields produces another line of debug output: calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) calendar-gui-Message: Check if default client matches (1137688431.10215.11@hex 1137688495.10224.6@hex) I have found no way of recovering from this problem short of nuking both my .evolution directory and the entire /apps/evolution schema in gconf, and then recreating my config from scratch. (Saving the gconf data when the problem is not occurring and then reloading the saved data does not work.) Steps to reproduce: 1. 2. 3. Actual results: Expected results: Does this happen every time? Other information: This occurs with every version of evolution, evolution-exchange, and evolution-data-server that I've tried. These include 2.5.3, 2.5.4, and 2.5.5.1 (1.5.3, 1.5.4, and 1.5.5 for EDS) as well as the "stable" 2.4.2.1 and friends. This is really frustrating; I'm getting rather tired of reentering all of my setup information, signatures, &etc every few days, and then waiting for Evolution to download all of my headers from the M sExchange server. If I could work effectively without calendaring and contacts, I'd probably have given up and gone back to Pine by now...
2.4.x had some issues, which are being fixed in 2.5.x. I have not observered such behavior with 2.5.5.1. Can you run evolution-exchange-storage from the separate console and paste the log messages when it fails to access addressbooks and calendar. Also, please set E2K_DEBUG=4 and run evolution-exchange-storage from that console and get the traces?
Will do. I'll report back once the problem has occurred again. Thanks!
a wild guess... does this happen after you open account editor? If that is the issue, it's fixed in 2.5.x.
Sorry to bust your guess... only for very large values of "after". :-) I'm running 2.5.5.1. It does happen "after" I use the account editor, 'cause I have to use that a few times to setup my accounts after nuking my settings... but it doesn't happen immediately after, it happens sometimes 2 or 3 days after. In any case, I came in this morning (having left Evo running) and the problem manifested. I'm attaching the logs from evolution-exchange-storage and evolution-data-storage, both run in their own console with E2K_DEBUG=4. I should note that my Exchange admin is willing to help debug in any way he can; let me know if any logs or settings from the server would be helpful.
Created attachment 58476 [details] tarball of logs from evolution-2.5.5.1, evolution-data-server-1.5.5, evolution-exchange-storage-2.5.5.1 bzipped tarball of debug data from evolution-data-server, and evolution-exchange-storage (3 files)
can you get the gconf source for one folder (say Contacts) when it works and when it stops working. (from ~/.gconf/apps/evolution/addressbook) you can change the user name and server name and paste it here.
Ok Since it's currently working, I've archived both the file you mentioned and the output of "gconftool-2 --dump /apps/evolution/addressbook". Once I see the bug manifest I'll upload the working and nonworking versions -- would the gconftool-2 output be preferable, or would you rather have the file? Thanks again for looking at this, Ole
gconftool-2 output is fine (for both addressbooks and calendar).
Ok, broken again. I'll attach the full gconftool-2 output for the addressbook in a moment (after running it through some trivial anonymizing) but diffing the two doesn't reveal much: diff -NauBb working/addressbook.dump b0rken/addressbook.dump --- working/addressbook.dump 2006-02-06 11:23:59.000000000 -0700 +++ b0rken/addressbook.dump 2006-02-06 18:11:33.000000000 -0700 @@ -37,6 +37,12 @@ </value> </entry> <entry> + <key>display/primary_addressbook</key> + <value> + <string>1138982262.1242.17@hex</string> + </value> + </entry> + <entry> <key>display/show_preview</key> <schema_key>/schemas/apps/evolution/addressbook/display/show_preview</schema_key> <value>
Created attachment 58875 [details] bzipped tarball of gconftool-2 output generated with "gconftool-2 --dump /apps/evolution/addressbook"
I should probably note that one of the first things I tried (which didn't work) was grabbing a copy of the gconf schema before the problem manifested and then overwriting the schema with a saved copy after the addressbook/calendar went autistic -- which didn't work. Evolution still ran the mail view fine, and the calendar and addressbook were still unable to utilize remote data. i.e. what I'd tried was: gconftool-2 --dump /apps/evolution > knowngood.xml (wait for problem to occur) gconftool-2 --recursive-unset /apps/evolution gconftool-2 --load knowngood.xml /apps/evolution (problem still occurs) Thanks, Ole
Thanks. Did you perform any operation, apart form reading/writing your mails/contacts/appointments, before it broke?
your gconf sources are fine and E2K_DEBUG traces also look fine. Can you run evolution-exchane-storage on a seperate console and see paste the log messages printed on the console, when addressbook and calendar folders stop working?
Also any error message popped up, when you try to access your address book?
(In reply to comment #12) > Thanks. > Did you perform any operation, apart form reading/writing your > mails/contacts/appointments, before it broke? > Just reading and sending email. My calendar is pretty empty at the moment. (which is why I usually first notice the problem with autocompletion of addresses.) I usually don't even stop/start evolution.
(In reply to comment #13) > your gconf sources are fine and E2K_DEBUG traces also look fine. > > Can you run evolution-exchane-storage on a seperate console and see paste the > log messages printed on the console, when addressbook and calendar folders > stop working? > I'm now running EDS, EES, and the frontend from three separate consoles, all with E2K_DEBUG=4, CAMEL_VERBOSE_DEBUG=1, and CAMEL_DEBUG=all. Contacts was dead again this morning, and I nuked my settings before I logged into bugzilla. I'll upload EES debug output as soon as I see the problem manifest again. The error messages I get: (In contacts view, when selecting GAL): Error loading addressbook. We were unable to open this addressbook. This either means you have entered an incorrect URI, or the server is unreachable. [OK] (In contacts view, when selecting remote contacts and searching): Unable to perform search. This query did not complete successfully. [OK] Debug output when I have it...
OK, something's just happened that never happened before and doesn't seem to make sense. Problem manifested sometime today, and it took me awhile to realize because I'd been replying to a lot of email without adding anyone to the recipients lists (so I hadn't observed when exactly the problem started up again.) I restarted all three progs so as to get only the error output, and decided not to run the output through "tee" but just to use a shell redirect. (I had been doing "evolution-exchange-storage 2>&1 | tee ~/log/ees-2.5.5.1.log" -- this time I just did "evolution-exchange-storage 2>&1 > ~/log/ees-2.5.5.1.log".) I thought this would let me write a line to the logs from another shell, which I could use as a separator (since evolution-exchange-storage seemed to be putting the entire contents of my inbox into the log -- I wanted to have something in the output to signify "startup has ended, cut from -here-.") I started up on the calendar page, which is where I'd shut down -- and lo and behold, my remote calendar was visible! And when I went to contacts, the GAL and remote contacts list were available again! I should note that before I broke down and posted the bug, each time it manifested I had restarted Evo a minimum of five or six times, fiddling with settings and trying to figure out what was going on. I've never had the problem appear and then go away again until now. In other news, I notice the unstable version available for download has revved to 2.5.90 -- should I build this and give it a shot, or is it simply a renaming of 2.5.5.1 for RC purposes? Thanks, Ole
Hi, So the problem has disappeared now? If it occurs again, please check that e-d-s has crashed or running? and Evo 2.5.90 and connector 2.5.9.0 are the latest versions and not just renamed version of 2.5.5.1. You can give it a try or if everyhting is woring fine... you can wait for 2.6.
new info: I believe it may have to do with the saved password setting. When I leave the "remember this password" checkbox blank, the problem fails to recur (over a period of 10 days, which is longer than I've ever stayed up before); when I check it, the problem eventually occurs again. So if I accidentally check that button, I guess I can work around it by changing the password on the Outlook server and forcing the authentication dialog to appear. Still a pain, but much better than re-entering all of my preferences...
I will try to get it tested with remember password, though i am not able to find any relation with remember password and error loading the folders, even after re-authenticating. if you try forget password (File->forget passwords) and restart evolution, e-d-s and exchange storage, does it load contacts and calendars for you?
Closing this bug report as no further information has been provided. Please feel free to reopen this bug if you can provide the information Sushma asked for. Thanks!
The original problem was so painful that once I found a workaround (leaving the password unsaved between evo sessions) I stuck with it. However, I've now finally gotten around to trying to figure out why I never get calendar alarms anymore, and come to discover bug #338909 -- which requires me to save my !(^%!%! password in order to get alarms! Holy catch-22 crapf*ck! Gah. I'll reluctantly reenable saving my password, and see what happens. (Note that in the intervening year the problem has not resurfaced while I've kept my password unsaved between sessions. Admittedly, this was over several release iterations.) Leaving RESOLVED INCOMPLETE until I reproduce the damn thing.