Bug 1030629

Thanks for a bug report. I believe it is related to email accounts which you have configured. For example, I have enabled about 5 accounts, and I do not see any such behaviour when I press Send/Receive button from the tool bar.

Could you install debuginfo packages for evolution-data-server and evolution, and maybe other evolution packages you might have installed and using, and grab a backtrace of evolution in the "waiting and CPU consuming" state, please? You can do that with this command:
   $ gdb --batch --ex "t a a bt" -pid=`pidof evolution` &>bt.txt
and attach the bt.txt file here. Please make sure you'll not expose any private data in it, like passwords or email addresses. I usually search for "pass" (quotes for clarity only).

By the way, is UI frozen, when the CPU is high, or you can click for example on the File menu and it'll expand with its submenus immediately?

Created attachment 825293 [details]
backtrace requested but it has problems!

First the problem does not happen on "send/receive". It happens on pressing "New". It also does not happen when replying! It is only when you press the "New" button for a new mail (or File>New>Mail Message).

When the high CPU occurs evolution is frozen, no response from any element of the UI, although I get messages saying that the aprticular window is not responding and asking if I want to wait or force close evolution.

Another potential help is that (I think) the first time you use the "New" button after a reboot it works fine. It is the second time that it takes forever. This appears to be true even if evolution has been stopped and restarted. I have not tried enough reboots to be certain of these to say they are facts but my current tests would indicate that this is the case.

Final note. I appear to have run into a bug related to #972351 in that backtrace is not working. Bear in mind that I am a novice at this and I just cut and pasted your command to extract the attached report. 

Let me know if there is anything else you need. I will respond but it may take a couple of days as I have a few things to do this week. It is a real pain when work gets in the way :-)

Thanks for the update. The backtrace looks fine, it shows some activity in DConf, notifying about changes. There can happen some ping-pong action with these updates, though I didn't see them when creating a new message, but rather in Edit->Preferences, thus what you see might be different from that what I have on mind [1], because the fix from there is already included in your 3.10.1.

Could you run this command on a terminal:
   $ gsettings monitor org.gnome.evolution.mail
and then create the new message, please? I tried it here, but it didn't show any activity, from which I suppose there was no change for me, but maybe there will be some for you.

[1] https://bugzilla.gnome.org/show_bug.cgi?id=698275

Milan,

Thanks for continuing to look at my problem. I appreciate that is is difficult to trap bugs without being able to reproduce them.

I have tired a few times:

After gsettings command, start Evolution and attempt to send a new mail:

[~]$ gsettings monitor org.gnome.evolution.mail
junk-default-plugin: ''
junk-default-plugin: ''
browser-close-on-reply-policy: 'always'
forward-style-name: 'inline'
show-headers: [('From', true), ('Reply-To', true), ('To', true), ('Cc', true), ('Bcc', true), ('Subject', true), ('Date', true), ('Newsgroups', true), ('Face', true), ('x-evolution-mailer', false)]
reply-style-name: 'quoted'
image-loading-policy: 'never'


The above output is just evolution starting. After about 5-10 minutes the new mail window appeared, at about the same time as a SIGSEGV occured (see below).

Send a new mail (No Output)

[~]$ gsettings monitor org.gnome.evolution.mail

This was repeated several times, each taking either 5-10 minutes to appear with the same SIGSEGV error happening at times. 

Today I have not managed to send a new mail! Either the SIGSEGV error or things just taking so long in the new mail window I abandoned the effort.

Replay to a mail(this always works without problem)
[~]$ gsettings monitor org.gnome.evolution.mail
browser-close-on-reply-policy: 'always'

NB while investigating this problem I have had several aborts (SIGSEGV) that I have reported as https://bugzilla.redhat.com/show_bug.cgi?id=1032690

Is it possible that this is a related problem?

What am I doing now? I built the problem system as a new install from the Fedora 20 alpha release live disc, installing extra software using yum. The data was then restored from a backup of my Fedora 19 system. I restored the mail system from a backup. I then made some changes as I have altered my mail systems since the backup! 

I am ditching that and reconfiguing all my accounts from scratch. I will let you know if that alters anything. Currently I have partially reconfigured just one account (the SMTP is not what I want at present, it will not work everywhere I work) and the new mail button responds as expected. I have this and the original configuration backed up.

I have just completed reconfiguring everything. It all looks good to me and new mail button responds within seconds, everytime.

It looks like I have an error in my backup! Not too sure what is going on. I have another machine with a similar configuration that restored without problems. I don't need the back-ups as my IMAP server ends up with a copy of everything.

I have even tried backing up and restoring Evolution a few times and it all appears to work OK.

Not only that but I have not seen the problem described in #1032690 since I cleared down my settings and started again.

I will give it a few days but it looks like this can be closed as due to Sue (Stupid user error).

*** Bug 1032690 has been marked as a duplicate of this bug. ***

Good it works for you, bad it required a start from scratch. This might not be needed, the restore/upgrade should work flawlessly. I'd like to know what caused the issue here, maybe it was just some GSettings data storage being broken, but I understand that the backup holds your personal data, thus I do not want you to share it.

I'm closing this for now, as you suggested, even I do not believe you did anything wrong here (referring to Sue).

doesn't the referenced duplicate https://bugzilla.redhat.com/show_bug.cgi?id=1032690 provide the missing data?

btw: I ran into this bug (the duplicate) just now with evolution 3.10.4, but I am unable to reproduce it. I just opened a mailto: link from firefox multiple times to a already running instance of evolution. Evolution hang for multiple seconds unable to repaint its windows and crashed after a while.

I am experiencing more than 10 crashes per day (the duplicate #1032690). My reproducer is just "reading memo-list". The symptom "everything takes a lot of time" is also here.

$ gsettings monitor org.gnome.evolution.mail
browser-close-on-reply-policy: 'always'
browser-close-on-reply-policy: 'always'
browser-close-on-reply-policy: 'always'
browser-close-on-reply-policy: 'always'
...

The 'strace -p `pidof evolution`' gives me following: output
recvfrom(6, 0x20a70a4, 4096, 0, 0, 0)   = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=5, events=POLLIN}, {fd=6, events=POLLIN}, {fd=3, events=POLLIN}, {fd=12, events=POLLIN}], 4, 33) = 0 (Timeout)
recvfrom(6, 0x20a70a4, 4096, 0, 0, 0)   = -1 EAGAIN (Resource temporarily unavailable)
poll([{fd=5, events=POLLIN}, {fd=6, events=POLLIN}, {fd=3, events=POLLIN}, {fd=12, events=POLLIN}], 4, 33) = 0 (Timeout)
recvfrom(6, 0x20a70a4, 4096, 0, 0, 0)   = -1 EAGAIN (Resource temporarily unavailable)
...

During the "busy time", there is additionally:
===

close(21)                               = 0
futex(0x13a87d0, FUTEX_WAIT_PRIVATE, 2, NULL) = 0
futex(0x13a87d0, FUTEX_WAKE_PRIVATE, 1) = 0
futex(0x14713e0, FUTEX_WAIT_PRIVATE, 2, NULL) = 0
futex(0x14713e0, FUTEX_WAKE_PRIVATE, 1) = 0
futex(0x14158a0, FUTEX_WAIT_PRIVATE, 2, NULL) = -1 EAGAIN (Resource temporarily unavailable)
futex(0x14158a0, FUTEX_WAKE_PRIVATE, 1) = 0
futex(0x14158a0, FUTEX_WAIT_PRIVATE, 2, NULL) = -1 EAGAIN (Resource temporarily unavailable)
futex(0x14158a0, FUTEX_WAKE_PRIVATE, 1) = 0
futex(0x14158a0, FUTEX_WAKE_PRIVATE, 1) = 0
munmap(0x7f0ab7d32000, 40718)           = 0
munmap(0x7f0ab7d3c000, 1)               = 0
access("/run", F_OK)                    = 0
stat("/run", {st_mode=S_IFDIR|0755, st_size=1580, ...}) = 0
access("/run/user", F_OK)               = 0
stat("/run/user", {st_mode=S_IFDIR|0755, st_size=100, ...}) = 0
access("/run/user/21446", F_OK)         = 0
stat("/run/user/21446", {st_mode=S_IFDIR|0700, st_size=180, ...}) = 0
access("/run/user/21446/dconf", F_OK)   = 0
stat("/run/user/21446/dconf", {st_mode=S_IFDIR|0700, st_size=40, ...}) = 0
open("/run/user/21446/dconf/user", O_RDWR|O_CREAT, 0600) = 21
pwrite(21, "\0", 1, 1)                  = 1
mmap(NULL, 1, PROT_READ, MAP_SHARED, 21, 0) = 0x7f0ab7d3c000
close(21)                               = 0
open("/home/jmlich/.config/dconf/user", O_RDONLY) = 21
fstat(21, {st_mode=S_IFREG|0644, st_size=40718, ...}) = 0
mmap(NULL, 40718, PROT_READ, MAP_PRIVATE, 21, 0) = 0x7f0ab7d32000
close(21)                               = 0
munmap(0x7f0ab7d32000, 40718)           = 0
munmap(0x7f0ab7d3c000, 1)               = 0
access("/run", F_OK)                    = 0
stat("/run", {st_mode=S_IFDIR|0755, st_size=1580, ...}) = 0
access("/run/user", F_OK)               = 0
stat("/run/user", {st_mode=S_IFDIR|0755, st_size=100, ...}) = 0
access("/run/user/21446", F_OK)         = 0
stat("/run/user/21446", {st_mode=S_IFDIR|0700, st_size=180, ...}) = 0
access("/run/user/21446/dconf", F_OK)   = 0
stat("/run/user/21446/dconf", {st_mode=S_IFDIR|0700, st_size=40, ...}) = 0
open("/run/user/21446/dconf/user", O_RDWR|O_CREAT, 0600) = 21
pwrite(21, "\0", 1, 1)                  = 1
mmap(NULL, 1, PROT_READ, MAP_SHARED, 21, 0) = 0x7f0ab7d3c000
close(21)                               = 0
open("/home/jmlich/.config/dconf/user", O_RDONLY) = 21
fstat(21, {st_mode=S_IFREG|0644, st_size=40718, ...}) = 0
mmap(NULL, 40718, PROT_READ, MAP_PRIVATE, 21, 0) = 0x7f0ab7d32000
close(21)                               = 0
===


The backtrace is attached in duplicate bug. Let me know how I can provide you more details.

Ben Khan asked me yesterday about a similar delay when starting a new mail, and a backtrace revealed GtkFileChooser synchronously querying all his GVFS mounts in the main thread to assemble the "Places" sidebar, which locked up the UI.

May not be the same as what others are seeing here, but it's something else to look out for.

I suppose the crash from bug #1032690 is fixed upstream by bug:
https://bugzilla.gnome.org/show_bug.cgi?id=710367
I might mark it as a duplicate of it incorrectly, I'm sorry for that.

The "ping-pong" on false change notifications is fixed upstream in evolution with ending commit
https://git.gnome.org/browse/evolution/commit/?id=6e9e7b0676
for 3.13.3+, while a similar changes landed into evolution 3.12.3+ as well.

I forgot to mention, other reasons can influence the window open too, like what Matthew mentioned in the comment #10.