Re: [SyncEvolution] Strange empty events in myFUNAMBOL ... and timeouts
Hello Stefano!
Did you find out more about this, perhaps by looking at log files as I
suggested in my email below?
I've checked our nightly testing, looking for similar failures. I only
found quite a number of failures due to timeouts. Like other SyncML
clients, SyncEvolution resends a message if it doesn't get a response
from the SyncML server. Synthesis told us that Nokia phones do that
quite soon, so we picked the same timeout of 1 minute.
This seems to be too short for the Funambol server:
http://runtests.syncevolution.org/2009-12-08-22-00/head-evolution-testing...
http://runtests.syncevolution.org/2009-12-08-22-00/head-evolution-testing...
Is the load on the server really so high that it can't reply to messages
within a minute? This is specific to the Funambol server, others
(ScheduleWorld, Synthesis, ...) don't have this issue. So it's not
related to our network or our proxies.
Furthermore, the server then fails to handle the resent message. Is that
something that the server should be able to cope with? It must,
otherwise it probably also fails with Nokia phones under similar load
conditions.
Bye, Patrick
On Fri, 2009-12-04 at 13:29 +0100, Patrick Ohly wrote:
On Fri, 2009-12-04 at 11:17 +0000, Stefano Maffulli wrote:
> The weird thing is that in myFUNAMBOL table there are two events, but in
> SyncEvolution backup dirs I find only one. After a slow sync there are 5
> unnamed events in the database and 3 in the dir
> backup/SyncEvolution-funambol-2009-12-04-09-57/calendar.after
>
> We're investigating further, but we still haven't found a way to
> reproduce. Did anybody notice anything like this before?
No, never heard of this before. My recommendation is to enable a high
"maxlogdir" setting and then analyze the logs to figure out where the
events occurred for the first time.
On the other hand, you seem to have a situation where they are produced
already, so the log in backup/SyncEvolution-funambol-2009-12-04-09-57/
should tell you whether it was the client or the server who created the
new events.
> BTW, is there any way to count the total number of events stored in
> Evolution?
The statistics printed by SyncEvolution before and after a sync should
match the number of VEVENTs. You can also identify the calendar.ics file
("syncevolution" with no parameters) and the do a "grep BEGIN:VEVENT |
wc -l".